{"doc_id": "doc_000001", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-06-15 staffing section, Project Nereid listed Nora Bauer as the lead."}, {"sent_id": 3, "text": "A compute allocation for Project Nereid was logged on 2026-04-03."}, {"sent_id": 4, "text": "Node Birch-04 supplied 3 GPUs."}, {"sent_id": 5, "text": "The 2026-04-28 result for Project Sonata put Aster-3B on RiverBench-3 at 0.634 evidence F1, with Node Willow-05 handling the run."}, {"sent_id": 6, "text": "The 2026-05-09 method entry for Project Saffron moved from confidence calibration to chain verification."}, {"sent_id": 7, "text": "Milestone V1 for Project Meridian was revised on 2026-04-08."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-04 to 2026-05-14."}, {"sent_id": 9, "text": "Project Sonata ran Marble-7B on NereidNotes using Node Maple-01 on 2026-04-07."}, {"sent_id": 10, "text": "The reported evidence F1 was 0.856."}, {"sent_id": 11, "text": "River-3B was the model selected by Project Aster on 2026-06-07 for the evidence retrieval track."}], "source": "synthetic_rewritten", "title": "Run Brief 001"}
{"doc_id": "doc_000002", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-05-23, Project Nereid retired calibrated voting in favor of confidence calibration."}, {"sent_id": 3, "text": "Using Node Cedar-02, Orchid-3B produced 0.815 accuracy on LumenFacts for Project Anchor on 2026-06-27."}, {"sent_id": 4, "text": "Project Meridian received 2 GPUs from Node Birch-04 in the 2026-06-27 allocation log."}, {"sent_id": 5, "text": "For error analysis, the 2026-04-22 note says Project Sonata went with Marble-7B."}, {"sent_id": 6, "text": "Using Node Fir-10, Vela-8B produced 0.595 latency efficiency score on LumenFacts-3 for Project Saffron on 2026-04-28."}, {"sent_id": 7, "text": "Project Saffron had a method update on 2026-04-04."}, {"sent_id": 8, "text": "The old approach was chain verification, and the replacement was structured prompting."}], "source": "synthetic_rewritten", "title": "Review Packet 002"}
{"doc_id": "doc_000003", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-21 result for Project Sonata put Mira-7B on TraceEval-3 at 0.838 macro F1, with Node Sycamore-13 handling the run."}, {"sent_id": 3, "text": "Project Saffron received 2 GPUs from Node Willow-05 in the 2026-05-01 allocation log."}, {"sent_id": 4, "text": "Milestone D1 for Project Meridian was revised on 2026-06-04."}, {"sent_id": 5, "text": "The schedule moved from 2026-06-16 to 2026-06-24."}, {"sent_id": 6, "text": "Project Saffron recorded a model choice for error analysis on 2026-04-12."}, {"sent_id": 7, "text": "The chosen model was Marble-3B."}, {"sent_id": 8, "text": "Milestone D1 for Project Nereid was revised on 2026-04-17."}, {"sent_id": 9, "text": "The schedule moved from 2026-04-24 to 2026-04-23."}, {"sent_id": 10, "text": "The review note for Amber Ranking Study 1 was dated 2026-04-06."}, {"sent_id": 11, "text": "It described the method as confidence calibration and said it did not use a reward model."}, {"sent_id": 12, "text": "The staffing note for Project Sonata was dated 2026-06-23."}, {"sent_id": 13, "text": "It named Samir Kwan for the retrieval owner duties."}, {"sent_id": 14, "text": "Project Saffron recorded a model choice for reranking on 2026-06-19."}, {"sent_id": 15, "text": "The chosen model was Vela-7B."}], "source": "synthetic_rewritten", "title": "Run Brief 003"}
{"doc_id": "doc_000004", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone N1 for Project Anchor was revised on 2026-05-07."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-29 to 2026-06-08."}, {"sent_id": 4, "text": "Mira-7B was the model selected by Project Sonata on 2026-06-24 for the calibration track."}, {"sent_id": 5, "text": "Aster-8B was the model selected by Project Sonata on 2026-05-29 for the evidence retrieval track."}, {"sent_id": 6, "text": "Project Meridian ran Cedar-3B on VestaLogs-2 using Node Laurel-11 on 2026-06-20."}, {"sent_id": 7, "text": "The reported accuracy was 0.681."}, {"sent_id": 8, "text": "The 2026-06-30 result for Project Saffron put Marble-3B on VestaLogs-3 at 0.631 latency efficiency score, with Node Maple-01 handling the run."}, {"sent_id": 9, "text": "Project Anchor's Milestone B1 deadline change on 2026-06-15 replaced 2026-07-01 with 2026-07-09."}, {"sent_id": 10, "text": "Project Aster's Milestone N1 deadline change on 2026-06-19 replaced 2026-07-06 with 2026-07-11."}, {"sent_id": 11, "text": "A compute allocation for Project Aster was logged on 2026-04-23."}, {"sent_id": 12, "text": "Node Spruce-03 supplied 3 GPUs."}], "source": "synthetic_rewritten", "title": "Review Packet 004"}
{"doc_id": "doc_000005", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian logged a personnel update on 2026-04-26: Elian Shah took the data steward responsibilities."}, {"sent_id": 3, "text": "On 2026-06-07, Project Meridian retired structured prompting in favor of QLoRA adaptation."}, {"sent_id": 4, "text": "Project Aster received 1 GPU from Node Spruce-03 in the 2026-05-30 allocation log."}, {"sent_id": 5, "text": "The 2026-04-28 result for Project Saffron put Mira-7B on MemoTrace-2 at 0.673 accuracy, with Node Sycamore-13 handling the run."}, {"sent_id": 6, "text": "For evidence retrieval, the 2026-06-19 note says Project Meridian went with Atlas-3B."}, {"sent_id": 7, "text": "The 2026-04-22 method entry for Project Sonata moved from threshold search to reward reranking."}, {"sent_id": 8, "text": "A compute allocation for Project Nereid was logged on 2026-05-23."}, {"sent_id": 9, "text": "Node Birch-04 supplied 3 GPUs."}, {"sent_id": 10, "text": "Project Anchor recorded a model choice for reranking on 2026-06-26."}, {"sent_id": 11, "text": "The chosen model was Orchid-7B."}, {"sent_id": 12, "text": "Project Sonata recorded a model choice for claim classification on 2026-05-17."}, {"sent_id": 13, "text": "The chosen model was Quartz-7B."}], "source": "synthetic_rewritten", "title": "Run Brief 005"}
{"doc_id": "doc_000006", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid received 1 GPU from Node Elm-08 in the 2026-04-04 allocation log."}, {"sent_id": 3, "text": "Mira-3B was the model selected by Project Saffron on 2026-06-17 for the calibration track."}, {"sent_id": 4, "text": "Project Saffron ran Vela-3B on LumenFacts-3 using Node Fir-10 on 2026-06-20."}, {"sent_id": 5, "text": "The reported accuracy was 0.655."}, {"sent_id": 6, "text": "The 2026-06-11 failure note says Aster-8B on Node Willow-05 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 7, "text": "A compute allocation for Project Aster was logged on 2026-04-10."}, {"sent_id": 8, "text": "Node Juniper-06 supplied 3 GPUs."}, {"sent_id": 9, "text": "Project Nereid received 3 GPUs from Node Elm-08 in the 2026-04-25 allocation log."}, {"sent_id": 10, "text": "The 2026-06-28 method entry for Project Nereid moved from sentence pruning to cross-encoder reranking."}], "source": "synthetic_rewritten", "title": "Review Packet 006"}
{"doc_id": "doc_000007", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Saffron received 4 GPUs from Node Fir-10 in the 2026-04-18 allocation log."}, {"sent_id": 3, "text": "For claim classification, the 2026-05-26 note says Project Meridian went with Kestrel-8B."}, {"sent_id": 4, "text": "Atlas-8B was the model selected by Project Nereid on 2026-04-16 for the evidence retrieval track."}, {"sent_id": 5, "text": "The 2026-04-11 compute note links Node Rowan-09 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 6, "text": "Project Sonata had a failed run on 2026-06-11 involving Mira-7B and Node Sycamore-13."}, {"sent_id": 7, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 8, "text": "Project Anchor had a failed run on 2026-05-08 involving Lumen-7B and Node Spruce-03."}, {"sent_id": 9, "text": "The recorded cause was a missing-index error."}, {"sent_id": 10, "text": "Lumen-7B was the model selected by Project Anchor on 2026-06-04 for the calibration track."}, {"sent_id": 11, "text": "A compute allocation for Project Nereid was logged on 2026-06-12."}, {"sent_id": 12, "text": "Node Aspen-01 supplied 3 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 007"}
{"doc_id": "doc_000008", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata ran Mira-7B on TraceEval-3 using Node Sycamore-13 on 2026-06-02."}, {"sent_id": 3, "text": "The reported macro F1 was 0.628."}, {"sent_id": 4, "text": "The staffing note for Project Meridian was dated 2026-04-13."}, {"sent_id": 5, "text": "It named Nadia Singh for the data steward duties."}, {"sent_id": 6, "text": "The 2026-04-03 failure note says Helix-8B on Node Poplar-12 failed for Project Anchor because of a missing-index error."}, {"sent_id": 7, "text": "A compute allocation for Project Saffron was logged on 2026-04-13."}, {"sent_id": 8, "text": "Node Pine-07 supplied 4 GPUs."}, {"sent_id": 9, "text": "Project Aster recorded a model choice for error analysis on 2026-04-17."}, {"sent_id": 10, "text": "The chosen model was Finch-7B."}, {"sent_id": 11, "text": "The 2026-05-28 failure note says Nova-8B on Node Aspen-01 failed for Project Meridian because of a missing-index error."}, {"sent_id": 12, "text": "The 2026-06-19 compute note links Node Rowan-09 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 13, "text": "On 2026-04-26, Project Sonata retired chain verification in favor of structured prompting."}], "source": "synthetic_rewritten", "title": "Review Packet 008"}
{"doc_id": "doc_000009", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-06-03, Project Aster shifted Milestone Z1: 2026-06-16 became 2026-06-21."}, {"sent_id": 3, "text": "Project Aster's Milestone N1 deadline change on 2026-06-03 replaced 2026-06-10 with 2026-06-09."}, {"sent_id": 4, "text": "Project Aster received 1 GPU from Node Cedar-02 in the 2026-05-21 allocation log."}, {"sent_id": 5, "text": "The review note for Quiet Retriever Study 3 was dated 2026-06-29."}, {"sent_id": 6, "text": "It described the method as sentence pruning and said it did not use a reward model."}, {"sent_id": 7, "text": "The 2026-04-18 compute note links Node Elm-08 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 8, "text": "Orchid-7B was the model selected by Project Anchor on 2026-06-14 for the reranking track."}, {"sent_id": 9, "text": "Project Sonata logged a personnel update on 2026-06-01: Mara Quinn took the lead responsibilities."}, {"sent_id": 10, "text": "Project Meridian's Milestone J1 deadline change on 2026-06-23 replaced 2026-07-13 with 2026-07-17."}, {"sent_id": 11, "text": "For Project Nereid, the run using Cedar-7B on Node Laurel-11 did not complete on 2026-04-24; the issue was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Run Brief 009"}
{"doc_id": "doc_000010", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Aster logged a personnel update on 2026-04-06: Felix Lind took the lead responsibilities."}, {"sent_id": 3, "text": "Project Saffron had a failed run on 2026-06-11 involving Aster-3B and Node Willow-05."}, {"sent_id": 4, "text": "The recorded cause was a missing-index error."}, {"sent_id": 5, "text": "On 2026-06-27, Project Sonata retired teacher distillation in favor of query rewriting."}, {"sent_id": 6, "text": "The staffing note for Project Aster was dated 2026-05-11."}, {"sent_id": 7, "text": "It named Dr. Noah Chen for the retrieval owner duties."}, {"sent_id": 8, "text": "On 2026-06-28, Project Sonata retired calibrated voting in favor of confidence calibration."}, {"sent_id": 9, "text": "For claim classification, the 2026-05-07 note says Project Anchor went with Helix-7B."}], "source": "synthetic_rewritten", "title": "Lab Digest 010"}
{"doc_id": "doc_000011", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Nereid ran Nova-7B on LabQA using Node Aspen-01 on 2026-06-20."}, {"sent_id": 3, "text": "The reported macro F1 was 0.720."}, {"sent_id": 4, "text": "The staffing note for Project Nereid was dated 2026-06-08."}, {"sent_id": 5, "text": "It named Dr. Julian Gray for the lead duties."}, {"sent_id": 6, "text": "On 2026-06-21, Project Saffron retired late interaction in favor of BM25 retrieval."}, {"sent_id": 7, "text": "In the 2026-06-16 staffing section, Project Saffron listed Lena Sol as the evaluation owner."}, {"sent_id": 8, "text": "The 2026-04-20 result for Project Aster put Orchid-3B on LabQA-3 at 0.830 macro F1, with Node Cedar-02 handling the run."}, {"sent_id": 9, "text": "The 2026-05-17 method entry for Project Anchor moved from dense retrieval to rank fusion."}], "source": "synthetic_rewritten", "title": "Project Update 011"}
{"doc_id": "doc_000012", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-20 compute note links Node Maple-01 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "Project Anchor logged a personnel update on 2026-04-19: Iris Lane took the evaluation owner responsibilities."}, {"sent_id": 4, "text": "The 2026-06-06 failure note says Mira-8B on Node Sycamore-13 failed for Project Saffron because of a checkpoint-mismatch error."}, {"sent_id": 5, "text": "On 2026-04-28, Project Meridian shifted Milestone D1: 2026-05-20 became 2026-05-26."}, {"sent_id": 6, "text": "Project Aster received 1 GPU from Node Poplar-12 in the 2026-04-11 allocation log."}, {"sent_id": 7, "text": "The staffing note for Project Nereid was dated 2026-04-20."}, {"sent_id": 8, "text": "It named Darian Hale for the retrieval owner duties."}], "source": "synthetic_rewritten", "title": "Lab Digest 012"}
{"doc_id": "doc_000013", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Mira-3B was the model selected by Project Saffron on 2026-05-08 for the calibration track."}, {"sent_id": 3, "text": "On 2026-04-19, Project Saffron retired threshold search in favor of reward reranking."}, {"sent_id": 4, "text": "Project Saffron had a failed run on 2026-05-21 involving Vela-3B and Node Fir-10."}, {"sent_id": 5, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 6, "text": "Project Meridian had a failed run on 2026-06-05 involving Cedar-3B and Node Laurel-11."}, {"sent_id": 7, "text": "The recorded cause was a missing-index error."}, {"sent_id": 8, "text": "Nova-3B was the model selected by Project Meridian on 2026-05-06 for the reranking track."}, {"sent_id": 9, "text": "The 2026-05-09 compute note links Node Spruce-03 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 10, "text": "The staffing note for Project Sonata was dated 2026-06-01."}, {"sent_id": 11, "text": "It named Sofia Frost for the lead duties."}, {"sent_id": 12, "text": "The 2026-06-08 result for Project Meridian put Cedar-3B on VestaLogs-2 at 0.741 accuracy, with Node Laurel-11 handling the run."}], "source": "synthetic_rewritten", "title": "Project Update 013"}
{"doc_id": "doc_000014", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-08, Project Sonata shifted Milestone X1: 2026-04-25 became 2026-04-30."}, {"sent_id": 3, "text": "Project Nereid had a failed run on 2026-05-21 involving Atlas-7B and Node Hazel-14."}, {"sent_id": 4, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 5, "text": "In the 2026-05-25 staffing section, Project Nereid listed Darian Hale as the retrieval owner."}, {"sent_id": 6, "text": "Project Saffron ran Quartz-3B on CedarQA-2 using Node Pine-07 on 2026-06-27."}, {"sent_id": 7, "text": "The reported accuracy was 0.607."}, {"sent_id": 8, "text": "Marble-7B was the model selected by Project Sonata on 2026-06-16 for the error analysis track."}, {"sent_id": 9, "text": "Using Node Hazel-14, Atlas-8B produced 0.729 accuracy on SignalSet-3 for Project Meridian on 2026-05-18."}, {"sent_id": 10, "text": "Project Sonata logged a personnel update on 2026-05-18: Dr. Jonas Nolan took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 014"}
{"doc_id": "doc_000015", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Aster received 1 GPU from Node Juniper-06 in the 2026-04-29 allocation log."}, {"sent_id": 3, "text": "The 2026-05-29 failure note says Quartz-8B on Node Pine-07 failed for Project Saffron because of a missing-index error."}, {"sent_id": 4, "text": "In the 2026-04-19 staffing section, Project Nereid listed Dr. Julian Gray as the lead."}, {"sent_id": 5, "text": "Project Sonata's Milestone L1 deadline change on 2026-05-07 replaced 2026-05-18 with 2026-05-13."}, {"sent_id": 6, "text": "Project Nereid received 1 GPU from Node Birch-04 in the 2026-04-24 allocation log."}, {"sent_id": 7, "text": "The 2026-06-30 result for Project Aster put Finch-3B on NereidNotes-2 at 0.566 macro F1, with Node Juniper-06 handling the run."}], "source": "synthetic_rewritten", "title": "Project Update 015"}
{"doc_id": "doc_000016", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Using Node Maple-01, Marble-7B produced 0.571 accuracy on VestaLogs-3 for Project Saffron on 2026-06-29."}, {"sent_id": 3, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-06."}, {"sent_id": 4, "text": "The chosen model was Cedar-8B."}, {"sent_id": 5, "text": "Project Aster recorded a model choice for calibration on 2026-05-05."}, {"sent_id": 6, "text": "The chosen model was Lumen-3B."}, {"sent_id": 7, "text": "Project Aster recorded a model choice for reranking on 2026-06-02."}, {"sent_id": 8, "text": "The chosen model was Orchid-3B."}, {"sent_id": 9, "text": "On 2026-05-02, Project Sonata retired structured prompting in favor of QLoRA adaptation."}, {"sent_id": 10, "text": "Project Aster ran Helix-3B on OrionBench-2 using Node Poplar-12 on 2026-06-28."}, {"sent_id": 11, "text": "The reported evidence F1 was 0.602."}, {"sent_id": 12, "text": "Milestone R1 for Project Sonata was revised on 2026-04-03."}, {"sent_id": 13, "text": "The schedule moved from 2026-04-22 to 2026-04-19."}], "source": "synthetic_rewritten", "title": "Lab Digest 016"}
{"doc_id": "doc_000017", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-04-22 method entry for Project Meridian moved from chain verification to structured prompting."}, {"sent_id": 3, "text": "Project Sonata logged a personnel update on 2026-06-08: Samir Ames took the lead responsibilities."}, {"sent_id": 4, "text": "Project Sonata recorded a model choice for claim classification on 2026-05-14."}, {"sent_id": 5, "text": "The chosen model was Quartz-3B."}, {"sent_id": 6, "text": "For claim classification, the 2026-04-12 note says Project Meridian went with Kestrel-3B."}, {"sent_id": 7, "text": "The 2026-05-10 method entry for Project Sonata moved from LoRA adaptation to alias expansion."}, {"sent_id": 8, "text": "Project Meridian's Milestone V1 deadline change on 2026-05-28 replaced 2026-06-13 with 2026-06-21."}, {"sent_id": 9, "text": "A compute allocation for Project Saffron was logged on 2026-05-30."}, {"sent_id": 10, "text": "Node Pine-07 supplied 2 GPUs."}, {"sent_id": 11, "text": "The 2026-05-22 compute note links Node Maple-01 to Project Saffron with an allocation of 2 GPUs."}], "source": "synthetic_rewritten", "title": "Project Update 017"}
{"doc_id": "doc_000018", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-05-11 staffing section, Project Saffron listed Petra Gray as the evaluation owner."}, {"sent_id": 3, "text": "The staffing note for Project Saffron was dated 2026-04-26."}, {"sent_id": 4, "text": "It named Petra Gray for the evaluation owner duties."}, {"sent_id": 5, "text": "Helix-7B was the model selected by Project Aster on 2026-06-11 for the claim classification track."}, {"sent_id": 6, "text": "On 2026-04-05, Project Saffron retired query rewriting in favor of sentence pruning."}, {"sent_id": 7, "text": "The 2026-06-10 method entry for Project Saffron moved from contrastive tuning to threshold search."}, {"sent_id": 8, "text": "Project Sonata ran Mira-3B on TraceEval-3 using Node Sycamore-13 on 2026-06-15."}, {"sent_id": 9, "text": "The reported evidence F1 was 0.628."}, {"sent_id": 10, "text": "In the 2026-06-01 staffing section, Project Aster listed Mina Shah as the lead."}], "source": "synthetic_rewritten", "title": "Lab Digest 018"}
{"doc_id": "doc_000019", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For error analysis, the 2026-04-24 note says Project Aster went with Finch-3B."}, {"sent_id": 3, "text": "Project Nereid received 3 GPUs from Node Laurel-11 in the 2026-04-04 allocation log."}, {"sent_id": 4, "text": "Project Anchor ran River-8B on SignalSet-2 using Node Rowan-09 on 2026-04-28."}, {"sent_id": 5, "text": "The reported latency efficiency score was 0.569."}, {"sent_id": 6, "text": "The 2026-05-10 method entry for Project Aster moved from late interaction to BM25 retrieval."}, {"sent_id": 7, "text": "Marble-7B was the model selected by Project Saffron on 2026-06-17 for the error analysis track."}, {"sent_id": 8, "text": "On 2026-05-27, the Nested Verifier Study 4 review reported cross-encoder reranking and said it did not use a reward model."}, {"sent_id": 9, "text": "Project Nereid had a method update on 2026-05-03."}, {"sent_id": 10, "text": "The old approach was metric smoothing, and the replacement was hard-negative mining."}], "source": "synthetic_rewritten", "title": "Project Update 019"}
{"doc_id": "doc_000020", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for reranking on 2026-06-17."}, {"sent_id": 3, "text": "The chosen model was Nova-8B."}, {"sent_id": 4, "text": "Project Anchor ran River-3B on SignalSet-2 using Node Rowan-09 on 2026-04-06."}, {"sent_id": 5, "text": "The reported latency efficiency score was 0.809."}, {"sent_id": 6, "text": "Using Node Spruce-03, Lumen-3B produced 0.863 latency efficiency score on MemoTrace-3 for Project Anchor on 2026-05-12."}, {"sent_id": 7, "text": "For Project Sonata, the run using Vela-3B on Node Fir-10 did not complete on 2026-06-19; the issue was an out-of-memory error."}, {"sent_id": 8, "text": "Project Anchor ran Finch-3B on VestaLogs using Node Juniper-06 on 2026-06-23."}, {"sent_id": 9, "text": "The reported latency efficiency score was 0.731."}, {"sent_id": 10, "text": "Using Node Fir-10, Vela-7B produced 0.640 evidence F1 on LabQA-2 for Project Sonata on 2026-04-06."}, {"sent_id": 11, "text": "Pale Compass Study 2 appeared in a 2026-05-01 review entry using threshold search; the note said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Run Brief 020"}
{"doc_id": "doc_000021", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-13 compute note links Node Fir-10 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 3, "text": "Milestone B1 for Project Anchor was revised on 2026-06-16."}, {"sent_id": 4, "text": "The schedule moved from 2026-07-12 to 2026-07-22."}, {"sent_id": 5, "text": "Nova-8B was the model selected by Project Meridian on 2026-05-19 for the reranking track."}, {"sent_id": 6, "text": "The 2026-05-14 method entry for Project Anchor moved from temporal filtering to alias expansion."}, {"sent_id": 7, "text": "Project Aster's Milestone N1 deadline change on 2026-05-18 replaced 2026-06-04 with 2026-06-09."}, {"sent_id": 8, "text": "The 2026-04-03 compute note links Node Hazel-14 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 9, "text": "Project Meridian recorded a model choice for error analysis on 2026-06-25."}, {"sent_id": 10, "text": "The chosen model was Cedar-7B."}, {"sent_id": 11, "text": "Project Aster logged a personnel update on 2026-05-25: Anika Costa took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Review Packet 021"}
{"doc_id": "doc_000022", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "A compute allocation for Project Aster was logged on 2026-04-10."}, {"sent_id": 3, "text": "Node Rowan-09 supplied 1 GPU."}, {"sent_id": 4, "text": "For claim classification, the 2026-04-19 note says Project Meridian went with Kestrel-8B."}, {"sent_id": 5, "text": "Project Sonata's Milestone X1 deadline change on 2026-05-29 replaced 2026-06-15 with 2026-06-20."}, {"sent_id": 6, "text": "For Project Saffron, the run using Marble-7B on Node Maple-01 did not complete on 2026-06-26; the issue was a missing-index error."}, {"sent_id": 7, "text": "Project Aster ran Orchid-8B on LabQA-3 using Node Cedar-02 on 2026-04-27."}, {"sent_id": 8, "text": "The reported evidence F1 was 0.860."}, {"sent_id": 9, "text": "Milestone D2 for Project Saffron was revised on 2026-04-28."}, {"sent_id": 10, "text": "The schedule moved from 2026-05-18 to 2026-05-26."}, {"sent_id": 11, "text": "For claim classification, the 2026-04-09 note says Project Nereid went with Kestrel-7B."}, {"sent_id": 12, "text": "On 2026-04-05, Project Sonata retired reward reranking in favor of data mixing."}, {"sent_id": 13, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-05-13."}, {"sent_id": 14, "text": "The chosen model was Aster-8B."}], "source": "synthetic_rewritten", "title": "Run Brief 022"}
{"doc_id": "doc_000023", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-05-24."}, {"sent_id": 3, "text": "The chosen model was Atlas-8B."}, {"sent_id": 4, "text": "The 2026-06-28 method entry for Project Anchor moved from calibrated voting to confidence calibration."}, {"sent_id": 5, "text": "For Project Sonata, the run using Aster-8B on Node Willow-05 did not complete on 2026-04-17; the issue was an out-of-memory error."}, {"sent_id": 6, "text": "The 2026-04-17 failure note says Mira-3B on Node Sycamore-13 failed for Project Saffron because of a missing-index error."}, {"sent_id": 7, "text": "Project Meridian had a method update on 2026-04-12."}, {"sent_id": 8, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 9, "text": "Using Node Aspen-01, Nova-3B produced 0.705 accuracy on LumenFacts-2 for Project Meridian on 2026-05-02."}], "source": "synthetic_rewritten", "title": "Review Packet 023"}
{"doc_id": "doc_000024", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata had a failed run on 2026-06-25 involving Vela-3B and Node Fir-10."}, {"sent_id": 3, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 4, "text": "For Project Meridian, the run using Atlas-3B on Node Hazel-14 did not complete on 2026-05-14; the issue was a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-05-24 result for Project Meridian put Nova-3B on LumenFacts-2 at 0.855 latency efficiency score, with Node Aspen-01 handling the run."}, {"sent_id": 6, "text": "On 2026-06-05, Project Meridian shifted Milestone D1: 2026-06-27 became 2026-07-07."}, {"sent_id": 7, "text": "Using Node Aspen-01, Nova-8B produced 0.705 accuracy on LumenFacts-2 for Project Meridian on 2026-04-06."}, {"sent_id": 8, "text": "Project Sonata logged a personnel update on 2026-04-20: Sofia Frost took the lead responsibilities."}, {"sent_id": 9, "text": "A compute allocation for Project Anchor was logged on 2026-05-28."}, {"sent_id": 10, "text": "Node Juniper-06 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 024"}
{"doc_id": "doc_000025", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Nested Verifier Study 3 appeared in a 2026-06-22 review entry using rank fusion; the note said it did not use a reward model."}, {"sent_id": 3, "text": "For reranking, the 2026-06-18 note says Project Saffron went with Vela-8B."}, {"sent_id": 4, "text": "The 2026-06-02 result for Project Nereid put Nova-7B on LabQA at 0.810 macro F1, with Node Aspen-01 handling the run."}, {"sent_id": 5, "text": "The 2026-06-11 failure note says Orchid-7B on Node Cedar-02 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 6, "text": "Project Aster had a method update on 2026-04-26."}, {"sent_id": 7, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 8, "text": "Project Meridian recorded a model choice for error analysis on 2026-05-06."}, {"sent_id": 9, "text": "The chosen model was Cedar-3B."}, {"sent_id": 10, "text": "Project Aster ran Finch-3B on NereidNotes-2 using Node Juniper-06 on 2026-06-15."}, {"sent_id": 11, "text": "The reported evidence F1 was 0.836."}, {"sent_id": 12, "text": "Project Aster recorded a model choice for evidence retrieval on 2026-05-24."}, {"sent_id": 13, "text": "The chosen model was River-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 025"}
{"doc_id": "doc_000026", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-29 compute note links Node Juniper-06 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 3, "text": "For calibration, the 2026-05-03 note says Project Saffron went with Mira-8B."}, {"sent_id": 4, "text": "The staffing note for Project Saffron was dated 2026-06-14."}, {"sent_id": 5, "text": "It named Priya Vale for the evaluation owner duties."}, {"sent_id": 6, "text": "Project Aster had a failed run on 2026-05-09 involving Finch-7B and Node Juniper-06."}, {"sent_id": 7, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 8, "text": "On 2026-04-01, Project Sonata retired chain verification in favor of structured prompting."}, {"sent_id": 9, "text": "The 2026-04-25 method entry for Project Meridian moved from late interaction to BM25 retrieval."}, {"sent_id": 10, "text": "The 2026-05-02 compute note links Node Birch-04 to Project Meridian with an allocation of 4 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 026"}
{"doc_id": "doc_000027", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For error analysis, the 2026-05-03 note says Project Anchor went with Finch-8B."}, {"sent_id": 3, "text": "The review note for Stable Chains Study 3 was dated 2026-06-19."}, {"sent_id": 4, "text": "It described the method as BM25 retrieval and said it did not use a reward model."}, {"sent_id": 5, "text": "The 2026-06-15 compute note links Node Cedar-02 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 6, "text": "Project Sonata logged a personnel update on 2026-04-14: Rohan Park took the lead responsibilities."}, {"sent_id": 7, "text": "Project Nereid logged a personnel update on 2026-04-27: Evan Moss took the lead responsibilities."}, {"sent_id": 8, "text": "On 2026-06-03, Project Nereid shifted Milestone V1: 2026-06-14 became 2026-06-09."}, {"sent_id": 9, "text": "Project Sonata ran Marble-7B on NereidNotes using Node Maple-01 on 2026-06-28."}, {"sent_id": 10, "text": "The reported macro F1 was 0.706."}], "source": "synthetic_rewritten", "title": "Review Packet 027"}
{"doc_id": "doc_000028", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-05-07, Project Saffron shifted Milestone F1: 2026-05-25 became 2026-05-31."}, {"sent_id": 3, "text": "For Project Nereid, the run using Cedar-7B on Node Laurel-11 did not complete on 2026-06-06; the issue was an unstable-validation-loss error."}, {"sent_id": 4, "text": "The review note for Pale Compass Study 4 was dated 2026-06-05."}, {"sent_id": 5, "text": "It described the method as reward reranking and said it did not use a reward model."}, {"sent_id": 6, "text": "Milestone F1 for Project Sonata was revised on 2026-06-12."}, {"sent_id": 7, "text": "The schedule moved from 2026-07-05 to 2026-07-03."}, {"sent_id": 8, "text": "Project Aster logged a personnel update on 2026-04-21: Mina Shah took the lead responsibilities."}, {"sent_id": 9, "text": "The staffing note for Project Saffron was dated 2026-04-21."}, {"sent_id": 10, "text": "It named Owen Torres for the evaluation owner duties."}, {"sent_id": 11, "text": "Project Anchor had a method update on 2026-05-09."}, {"sent_id": 12, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 13, "text": "A compute allocation for Project Meridian was logged on 2026-05-30."}, {"sent_id": 14, "text": "Node Elm-08 supplied 2 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 028"}
{"doc_id": "doc_000029", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-19, Project Anchor retired LoRA adaptation in favor of alias expansion."}, {"sent_id": 3, "text": "For Project Aster, the run using Helix-8B on Node Poplar-12 did not complete on 2026-05-08; the issue was an unstable-validation-loss error."}, {"sent_id": 4, "text": "Project Meridian logged a personnel update on 2026-05-12: Elian Ames took the evaluation owner responsibilities."}, {"sent_id": 5, "text": "Mira-7B was the model selected by Project Sonata on 2026-06-26 for the calibration track."}, {"sent_id": 6, "text": "The 2026-04-24 failure note says Finch-8B on Node Juniper-06 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 7, "text": "On 2026-04-12, Project Aster retired hard-negative mining in favor of LoRA adaptation."}], "source": "synthetic_rewritten", "title": "Review Packet 029"}
{"doc_id": "doc_000030", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For calibration, the 2026-06-24 note says Project Saffron went with Mira-8B."}, {"sent_id": 3, "text": "Milestone N1 for Project Anchor was revised on 2026-06-30."}, {"sent_id": 4, "text": "The schedule moved from 2026-07-12 to 2026-07-20."}, {"sent_id": 5, "text": "On 2026-06-03, Project Anchor retired metric smoothing in favor of hard-negative mining."}, {"sent_id": 6, "text": "Project Meridian had a method update on 2026-04-23."}, {"sent_id": 7, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 8, "text": "For Project Meridian, the run using Atlas-7B on Node Hazel-14 did not complete on 2026-05-22; the issue was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "On 2026-05-05, Project Anchor shifted Milestone H1: 2026-05-19 became 2026-05-29."}], "source": "synthetic_rewritten", "title": "Project Update 030"}
{"doc_id": "doc_000031", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Aster logged a personnel update on 2026-04-06: Anika Sato took the lead responsibilities."}, {"sent_id": 3, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-05-03."}, {"sent_id": 4, "text": "The chosen model was Atlas-8B."}, {"sent_id": 5, "text": "The 2026-04-11 result for Project Meridian put Nimbus-7B on MemoTrace at 0.693 accuracy, with Node Elm-08 handling the run."}, {"sent_id": 6, "text": "Milestone D1 for Project Nereid was revised on 2026-04-01."}, {"sent_id": 7, "text": "The schedule moved from 2026-04-18 to 2026-04-27."}, {"sent_id": 8, "text": "Quartz-3B was the model selected by Project Saffron on 2026-04-09 for the claim classification track."}, {"sent_id": 9, "text": "On 2026-04-19, Project Nereid retired data mixing in favor of contrastive tuning."}, {"sent_id": 10, "text": "The staffing note for Project Anchor was dated 2026-06-29."}, {"sent_id": 11, "text": "It named Dr. Arun Bauer for the evaluation owner duties."}, {"sent_id": 12, "text": "The 2026-05-29 failure note says Cedar-7B on Node Laurel-11 failed for Project Meridian because of a checkpoint-mismatch error."}, {"sent_id": 13, "text": "In the 2026-05-12 staffing section, Project Aster listed Mina Adler as the retrieval owner."}], "source": "synthetic_rewritten", "title": "Lab Digest 031"}
{"doc_id": "doc_000032", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Cedar-02, Orchid-3B produced 0.575 latency efficiency score on LumenFacts for Project Anchor on 2026-04-19."}, {"sent_id": 3, "text": "On 2026-05-25, Project Meridian shifted Milestone P1: 2026-06-12 became 2026-06-18."}, {"sent_id": 4, "text": "The staffing note for Project Saffron was dated 2026-05-18."}, {"sent_id": 5, "text": "It named Dr. Theo Grant for the data steward duties."}, {"sent_id": 6, "text": "In the 2026-06-02 staffing section, Project Nereid listed Dr. Julian Gray as the lead."}, {"sent_id": 7, "text": "The 2026-06-04 method entry for Project Meridian moved from query rewriting to sentence pruning."}, {"sent_id": 8, "text": "Project Meridian's Milestone J1 deadline change on 2026-05-07 replaced 2026-05-27 with 2026-05-31."}, {"sent_id": 9, "text": "On 2026-05-03, Project Anchor retired alias expansion in favor of dense retrieval."}], "source": "synthetic_rewritten", "title": "Project Update 032"}
{"doc_id": "doc_000033", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Using Node Poplar-12, Helix-3B produced 0.797 latency efficiency score on CedarQA-3 for Project Anchor on 2026-04-11."}, {"sent_id": 3, "text": "The 2026-04-02 compute note links Node Sycamore-13 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 4, "text": "The 2026-04-04 result for Project Aster put Finch-8B on NereidNotes-2 at 0.806 macro F1, with Node Juniper-06 handling the run."}, {"sent_id": 5, "text": "The 2026-05-30 compute note links Node Juniper-06 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 6, "text": "Project Sonata had a failed run on 2026-05-07 involving Vela-7B and Node Fir-10."}, {"sent_id": 7, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 8, "text": "For evidence retrieval, the 2026-05-07 note says Project Anchor went with River-8B."}, {"sent_id": 9, "text": "The 2026-06-16 result for Project Sonata put Vela-3B on LabQA-2 at 0.610 evidence F1, with Node Fir-10 handling the run."}, {"sent_id": 10, "text": "The 2026-06-12 failure note says Nimbus-3B on Node Elm-08 failed for Project Meridian because of a missing-index error."}, {"sent_id": 11, "text": "Project Nereid logged a personnel update on 2026-04-20: Evan Iyer took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 033"}
{"doc_id": "doc_000034", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-06-28 method entry for Project Meridian moved from cross-encoder reranking to late interaction."}, {"sent_id": 3, "text": "The staffing note for Project Saffron was dated 2026-06-30."}, {"sent_id": 4, "text": "It named Priya Moss for the data steward duties."}, {"sent_id": 5, "text": "The 2026-06-21 method entry for Project Nereid moved from cross-encoder reranking to late interaction."}, {"sent_id": 6, "text": "Project Aster's Milestone Z1 deadline change on 2026-06-09 replaced 2026-07-02 with 2026-06-30."}, {"sent_id": 7, "text": "For Project Anchor, the run using Orchid-8B on Node Cedar-02 did not complete on 2026-06-04; the issue was a missing-index error."}, {"sent_id": 8, "text": "A compute allocation for Project Saffron was logged on 2026-06-29."}, {"sent_id": 9, "text": "Node Fir-10 supplied 2 GPUs."}, {"sent_id": 10, "text": "On 2026-06-15, the Stable Chains Study 3 review reported query rewriting and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Project Update 034"}
{"doc_id": "doc_000035", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For calibration, the 2026-06-05 note says Project Nereid went with Nimbus-8B."}, {"sent_id": 3, "text": "Project Nereid recorded a model choice for claim classification on 2026-04-16."}, {"sent_id": 4, "text": "The chosen model was Kestrel-3B."}, {"sent_id": 5, "text": "Nimbus-7B was the model selected by Project Meridian on 2026-04-15 for the calibration track."}, {"sent_id": 6, "text": "For Project Meridian, the run using Kestrel-3B on Node Birch-04 did not complete on 2026-04-24; the issue was a missing-index error."}, {"sent_id": 7, "text": "Project Saffron received 4 GPUs from Node Pine-07 in the 2026-04-25 allocation log."}, {"sent_id": 8, "text": "Project Saffron ran Aster-8B on SignalSet using Node Willow-05 on 2026-04-18."}, {"sent_id": 9, "text": "The reported accuracy was 0.619."}], "source": "synthetic_rewritten", "title": "Lab Digest 035"}
{"doc_id": "doc_000036", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Sonata received 3 GPUs from Node Maple-01 in the 2026-04-25 allocation log."}, {"sent_id": 3, "text": "Project Nereid had a method update on 2026-04-16."}, {"sent_id": 4, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}, {"sent_id": 5, "text": "Milestone D1 for Project Meridian was revised on 2026-05-06."}, {"sent_id": 6, "text": "The schedule moved from 2026-05-18 to 2026-05-22."}, {"sent_id": 7, "text": "Project Aster had a failed run on 2026-05-01 involving Helix-3B and Node Poplar-12."}, {"sent_id": 8, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 9, "text": "Project Meridian ran Nova-7B on LumenFacts-2 using Node Aspen-01 on 2026-04-13."}, {"sent_id": 10, "text": "The reported latency efficiency score was 0.735."}, {"sent_id": 11, "text": "Helix-8B was the model selected by Project Aster on 2026-05-29 for the claim classification track."}, {"sent_id": 12, "text": "The 2026-05-01 failure note says Finch-7B on Node Juniper-06 failed for Project Anchor because of a missing-index error."}, {"sent_id": 13, "text": "The 2026-06-20 compute note links Node Rowan-09 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 14, "text": "On 2026-05-18, Project Sonata shifted Milestone R1: 2026-06-06 became 2026-06-03."}], "source": "synthetic_rewritten", "title": "Project Update 036"}
{"doc_id": "doc_000037", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The staffing note for Project Sonata was dated 2026-06-30."}, {"sent_id": 3, "text": "It named Dr. Jonas Nolan for the retrieval owner duties."}, {"sent_id": 4, "text": "For claim classification, the 2026-05-31 note says Project Meridian went with Kestrel-3B."}, {"sent_id": 5, "text": "On 2026-06-24, the Nested Verifier Study 1 review reported calibrated voting and said it did not use a reward model."}, {"sent_id": 6, "text": "The 2026-05-09 compute note links Node Pine-07 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 7, "text": "On 2026-06-25, Project Nereid retired structured prompting in favor of QLoRA adaptation."}, {"sent_id": 8, "text": "Using Node Juniper-06, Finch-3B produced 0.866 macro F1 on NereidNotes-2 for Project Aster on 2026-04-18."}, {"sent_id": 9, "text": "The 2026-05-29 failure note says Marble-3B on Node Maple-01 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 10, "text": "On 2026-04-22, Project Nereid shifted Milestone P1: 2026-05-05 became 2026-05-10."}], "source": "synthetic_rewritten", "title": "Lab Digest 037"}
{"doc_id": "doc_000038", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For claim classification, the 2026-04-01 note says Project Meridian went with Kestrel-8B."}, {"sent_id": 3, "text": "Milestone X1 for Project Saffron was revised on 2026-06-01."}, {"sent_id": 4, "text": "The schedule moved from 2026-06-13 to 2026-06-21."}, {"sent_id": 5, "text": "For reranking, the 2026-05-06 note says Project Anchor went with Orchid-7B."}, {"sent_id": 6, "text": "A compute allocation for Project Aster was logged on 2026-05-01."}, {"sent_id": 7, "text": "Node Spruce-03 supplied 3 GPUs."}, {"sent_id": 8, "text": "Using Node Maple-01, Marble-3B produced 0.541 latency efficiency score on VestaLogs-3 for Project Saffron on 2026-04-18."}, {"sent_id": 9, "text": "Project Sonata had a method update on 2026-06-03."}, {"sent_id": 10, "text": "The old approach was late interaction, and the replacement was BM25 retrieval."}, {"sent_id": 11, "text": "Milestone N1 for Project Aster was revised on 2026-04-03."}, {"sent_id": 12, "text": "The schedule moved from 2026-04-20 to 2026-04-29."}, {"sent_id": 13, "text": "A compute allocation for Project Meridian was logged on 2026-04-11."}, {"sent_id": 14, "text": "Node Aspen-01 supplied 4 GPUs."}, {"sent_id": 15, "text": "Project Meridian logged a personnel update on 2026-05-24: Mira Nolan took the evaluation owner responsibilities."}], "source": "synthetic_rewritten", "title": "Project Update 038"}
{"doc_id": "doc_000039", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-06-07 method entry for Project Sonata moved from cross-encoder reranking to late interaction."}, {"sent_id": 3, "text": "Using Node Maple-01, Marble-3B produced 0.631 accuracy on VestaLogs-3 for Project Saffron on 2026-05-09."}, {"sent_id": 4, "text": "For Project Aster, the run using Finch-7B on Node Juniper-06 did not complete on 2026-04-30; the issue was an unstable-validation-loss error."}, {"sent_id": 5, "text": "In the 2026-06-30 staffing section, Project Aster listed Dr. Noah Chen as the retrieval owner."}, {"sent_id": 6, "text": "In the 2026-06-30 staffing section, Project Sonata listed Rohan Park as the lead."}, {"sent_id": 7, "text": "Project Sonata recorded a model choice for reranking on 2026-05-20."}, {"sent_id": 8, "text": "The chosen model was Vela-3B."}, {"sent_id": 9, "text": "Using Node Laurel-11, Cedar-3B produced 0.816 evidence F1 on NereidNotes-3 for Project Nereid on 2026-04-28."}], "source": "synthetic_rewritten", "title": "Lab Digest 039"}
{"doc_id": "doc_000040", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For calibration, the 2026-06-21 note says Project Aster went with Lumen-7B."}, {"sent_id": 3, "text": "On 2026-06-18, Project Sonata retired cross-encoder reranking in favor of late interaction."}, {"sent_id": 4, "text": "The staffing note for Project Sonata was dated 2026-05-19."}, {"sent_id": 5, "text": "It named Samir Kwan for the retrieval owner duties."}, {"sent_id": 6, "text": "On 2026-06-07, Project Sonata retired contrastive tuning in favor of threshold search."}, {"sent_id": 7, "text": "For claim classification, the 2026-06-17 note says Project Sonata went with Quartz-3B."}, {"sent_id": 8, "text": "Project Nereid's Milestone B2 deadline change on 2026-05-06 replaced 2026-05-25 with 2026-05-22."}, {"sent_id": 9, "text": "In the 2026-06-08 staffing section, Project Aster listed Talia Marin as the lead."}], "source": "synthetic_rewritten", "title": "Review Packet 040"}
{"doc_id": "doc_000041", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The review note for Amber Ranking Study 3 was dated 2026-05-11."}, {"sent_id": 3, "text": "It described the method as late interaction and said it did not use a reward model."}, {"sent_id": 4, "text": "A compute allocation for Project Saffron was logged on 2026-06-06."}, {"sent_id": 5, "text": "Node Willow-05 supplied 2 GPUs."}, {"sent_id": 6, "text": "Project Aster's Milestone B1 deadline change on 2026-04-28 replaced 2026-05-09 with 2026-05-04."}, {"sent_id": 7, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-06-04."}, {"sent_id": 8, "text": "The chosen model was Atlas-3B."}, {"sent_id": 9, "text": "On 2026-04-13, Project Aster shifted Milestone T1: 2026-04-28 became 2026-04-24."}, {"sent_id": 10, "text": "The 2026-04-30 failure note says Kestrel-3B on Node Birch-04 failed for Project Meridian because of a missing-index error."}, {"sent_id": 11, "text": "Orchid-7B was the model selected by Project Anchor on 2026-04-12 for the reranking track."}, {"sent_id": 12, "text": "Milestone F1 for Project Saffron was revised on 2026-04-23."}, {"sent_id": 13, "text": "The schedule moved from 2026-05-11 to 2026-05-21."}], "source": "synthetic_rewritten", "title": "Run Brief 041"}
{"doc_id": "doc_000042", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For reranking, the 2026-05-06 note says Project Nereid went with Nova-7B."}, {"sent_id": 3, "text": "Milestone F1 for Project Sonata was revised on 2026-06-05."}, {"sent_id": 4, "text": "The schedule moved from 2026-06-18 to 2026-06-23."}, {"sent_id": 5, "text": "Project Sonata ran Vela-8B on LabQA-2 using Node Fir-10 on 2026-05-12."}, {"sent_id": 6, "text": "The reported macro F1 was 0.850."}, {"sent_id": 7, "text": "On 2026-06-04, Project Meridian shifted Milestone V1: 2026-06-30 became 2026-07-06."}, {"sent_id": 8, "text": "Using Node Pine-07, Quartz-7B produced 0.622 evidence F1 on OrionBench for Project Sonata on 2026-04-07."}, {"sent_id": 9, "text": "The 2026-06-10 method entry for Project Meridian moved from QLoRA adaptation to hybrid retrieval."}, {"sent_id": 10, "text": "The staffing note for Project Meridian was dated 2026-06-09."}, {"sent_id": 11, "text": "It named Dr. Ravi Reed for the evaluation owner duties."}, {"sent_id": 12, "text": "The staffing note for Project Aster was dated 2026-06-23."}, {"sent_id": 13, "text": "It named Talia Reed for the retrieval owner duties."}], "source": "synthetic_rewritten", "title": "Review Packet 042"}
{"doc_id": "doc_000043", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-28 method entry for Project Aster moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 3, "text": "On 2026-05-01, Project Sonata shifted Milestone L1: 2026-05-12 became 2026-05-07."}, {"sent_id": 4, "text": "Project Nereid had a method update on 2026-05-30."}, {"sent_id": 5, "text": "The old approach was confidence calibration, and the replacement was chain verification."}, {"sent_id": 6, "text": "Project Aster had a failed run on 2026-04-16 involving Lumen-3B and Node Spruce-03."}, {"sent_id": 7, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 8, "text": "Project Anchor recorded a model choice for claim classification on 2026-05-27."}, {"sent_id": 9, "text": "The chosen model was Helix-3B."}, {"sent_id": 10, "text": "Project Sonata logged a personnel update on 2026-05-18: Samir Ames took the lead responsibilities."}, {"sent_id": 11, "text": "Project Aster's Milestone B1 deadline change on 2026-06-12 replaced 2026-07-03 with 2026-07-08."}, {"sent_id": 12, "text": "The 2026-05-16 compute note links Node Maple-01 to Project Sonata with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Run Brief 043"}
{"doc_id": "doc_000044", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For Project Sonata, the run using Marble-8B on Node Maple-01 did not complete on 2026-05-21; the issue was an unstable-validation-loss error."}, {"sent_id": 3, "text": "For reranking, the 2026-06-19 note says Project Nereid went with Nova-8B."}, {"sent_id": 4, "text": "For reranking, the 2026-04-12 note says Project Meridian went with Nova-8B."}, {"sent_id": 5, "text": "Quiet Retriever Study 2 appeared in a 2026-04-29 review entry using evidence pooling; the note said it did not use a reward model."}, {"sent_id": 6, "text": "For reranking, the 2026-04-26 note says Project Sonata went with Vela-3B."}, {"sent_id": 7, "text": "Project Saffron had a method update on 2026-06-03."}, {"sent_id": 8, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 9, "text": "Project Nereid had a method update on 2026-05-24."}, {"sent_id": 10, "text": "The old approach was hard-negative mining, and the replacement was LoRA adaptation."}, {"sent_id": 11, "text": "The 2026-06-23 result for Project Saffron put Marble-7B on VestaLogs-3 at 0.601 accuracy, with Node Maple-01 handling the run."}], "source": "synthetic_rewritten", "title": "Review Packet 044"}
{"doc_id": "doc_000045", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian had a method update on 2026-05-13."}, {"sent_id": 3, "text": "The old approach was threshold search, and the replacement was reward reranking."}, {"sent_id": 4, "text": "Milestone D2 for Project Sonata was revised on 2026-04-17."}, {"sent_id": 5, "text": "The schedule moved from 2026-05-12 to 2026-05-17."}, {"sent_id": 6, "text": "Project Saffron had a failed run on 2026-06-18 involving Aster-8B and Node Willow-05."}, {"sent_id": 7, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 8, "text": "A compute allocation for Project Meridian was logged on 2026-05-09."}, {"sent_id": 9, "text": "Node Hazel-14 supplied 4 GPUs."}, {"sent_id": 10, "text": "Project Nereid received 1 GPU from Node Birch-04 in the 2026-04-02 allocation log."}, {"sent_id": 11, "text": "The 2026-04-23 method entry for Project Saffron moved from reward reranking to data mixing."}, {"sent_id": 12, "text": "Project Aster recorded a model choice for error analysis on 2026-06-14."}, {"sent_id": 13, "text": "The chosen model was Finch-7B."}, {"sent_id": 14, "text": "For Project Saffron, the run using Aster-7B on Node Willow-05 did not complete on 2026-05-02; the issue was a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 045"}
{"doc_id": "doc_000046", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-05-14, Project Saffron shifted Milestone R1: 2026-05-28 became 2026-06-03."}, {"sent_id": 3, "text": "On 2026-06-26, Project Anchor shifted Milestone H1: 2026-07-10 became 2026-07-20."}, {"sent_id": 4, "text": "Project Anchor recorded a model choice for reranking on 2026-05-31."}, {"sent_id": 5, "text": "The chosen model was Orchid-3B."}, {"sent_id": 6, "text": "Project Aster logged a personnel update on 2026-05-17: Anika Sato took the lead responsibilities."}, {"sent_id": 7, "text": "On 2026-04-09, Project Anchor shifted Milestone T1: 2026-04-19 became 2026-04-25."}, {"sent_id": 8, "text": "For reranking, the 2026-04-28 note says Project Sonata went with Vela-7B."}, {"sent_id": 9, "text": "For reranking, the 2026-05-24 note says Project Anchor went with Orchid-7B."}, {"sent_id": 10, "text": "The 2026-05-03 method entry for Project Saffron moved from document chunking to metric smoothing."}, {"sent_id": 11, "text": "On 2026-05-06, the Amber Ranking Study 4 review reported late interaction and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Review Packet 046"}
{"doc_id": "doc_000047", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-22, Project Sonata retired BM25 retrieval in favor of evidence pooling."}, {"sent_id": 3, "text": "The staffing note for Project Nereid was dated 2026-06-16."}, {"sent_id": 4, "text": "It named Evan Moss for the lead duties."}, {"sent_id": 5, "text": "On 2026-04-13, the Stable Chains Study 4 review reported alias expansion and said it did not use a reward model."}, {"sent_id": 6, "text": "For Project Anchor, the run using Helix-7B on Node Poplar-12 did not complete on 2026-04-10; the issue was a checkpoint-mismatch error."}, {"sent_id": 7, "text": "Using Node Laurel-11, Cedar-7B produced 0.741 accuracy on VestaLogs-2 for Project Meridian on 2026-04-04."}, {"sent_id": 8, "text": "Milestone Z1 for Project Anchor was revised on 2026-04-29."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-17 to 2026-05-27."}, {"sent_id": 10, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-06-14."}, {"sent_id": 11, "text": "The chosen model was Atlas-3B."}], "source": "synthetic_rewritten", "title": "Run Brief 047"}
{"doc_id": "doc_000048", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For Project Anchor, the run using River-7B on Node Rowan-09 did not complete on 2026-04-17; the issue was a checkpoint-mismatch error."}, {"sent_id": 3, "text": "Project Meridian received 4 GPUs from Node Birch-04 in the 2026-04-18 allocation log."}, {"sent_id": 4, "text": "Milestone H1 for Project Aster was revised on 2026-04-30."}, {"sent_id": 5, "text": "The schedule moved from 2026-05-19 to 2026-05-16."}, {"sent_id": 6, "text": "Vela-7B was the model selected by Project Saffron on 2026-04-08 for the reranking track."}, {"sent_id": 7, "text": "For error analysis, the 2026-06-21 note says Project Aster went with Finch-3B."}, {"sent_id": 8, "text": "Project Anchor recorded a model choice for calibration on 2026-06-19."}, {"sent_id": 9, "text": "The chosen model was Lumen-7B."}, {"sent_id": 10, "text": "Milestone F1 for Project Saffron was revised on 2026-04-08."}, {"sent_id": 11, "text": "The schedule moved from 2026-04-26 to 2026-05-06."}], "source": "synthetic_rewritten", "title": "Review Packet 048"}
{"doc_id": "doc_000049", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Nimbus-7B was the model selected by Project Meridian on 2026-04-21 for the calibration track."}, {"sent_id": 3, "text": "Cedar-8B was the model selected by Project Meridian on 2026-04-07 for the error analysis track."}, {"sent_id": 4, "text": "The staffing note for Project Meridian was dated 2026-06-22."}, {"sent_id": 5, "text": "It named Nadia Chen for the evaluation owner duties."}, {"sent_id": 6, "text": "On 2026-06-25, Project Anchor retired confidence calibration in favor of chain verification."}, {"sent_id": 7, "text": "Project Meridian ran Nimbus-7B on MemoTrace using Node Elm-08 on 2026-05-12."}, {"sent_id": 8, "text": "The reported accuracy was 0.603."}, {"sent_id": 9, "text": "Project Meridian had a method update on 2026-04-11."}, {"sent_id": 10, "text": "The old approach was QLoRA adaptation, and the replacement was hybrid retrieval."}, {"sent_id": 11, "text": "On 2026-05-21, Project Saffron shifted Milestone R1: 2026-06-14 became 2026-06-18."}, {"sent_id": 12, "text": "Project Sonata ran Vela-7B on LabQA-2 using Node Fir-10 on 2026-06-09."}, {"sent_id": 13, "text": "The reported macro F1 was 0.580."}], "source": "synthetic_rewritten", "title": "Run Brief 049"}
{"doc_id": "doc_000050", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-05 result for Project Meridian put Atlas-3B on SignalSet-3 at 0.879 latency efficiency score, with Node Hazel-14 handling the run."}, {"sent_id": 3, "text": "Project Aster logged a personnel update on 2026-04-21: Felix Brooks took the retrieval owner responsibilities."}, {"sent_id": 4, "text": "Project Nereid had a failed run on 2026-04-09 involving Nova-7B and Node Aspen-01."}, {"sent_id": 5, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 6, "text": "For Project Meridian, the run using Nova-7B on Node Aspen-01 did not complete on 2026-05-23; the issue was a checkpoint-mismatch error."}, {"sent_id": 7, "text": "A compute allocation for Project Saffron was logged on 2026-06-22."}, {"sent_id": 8, "text": "Node Fir-10 supplied 4 GPUs."}, {"sent_id": 9, "text": "The 2026-05-16 method entry for Project Meridian moved from alias expansion to dense retrieval."}, {"sent_id": 10, "text": "A compute allocation for Project Nereid was logged on 2026-04-11."}, {"sent_id": 11, "text": "Node Hazel-14 supplied 3 GPUs."}, {"sent_id": 12, "text": "Using Node Juniper-06, Finch-7B produced 0.536 evidence F1 on NereidNotes-2 for Project Aster on 2026-06-23."}, {"sent_id": 13, "text": "The 2026-06-24 compute note links Node Elm-08 to Project Nereid with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Lab Digest 050"}
{"doc_id": "doc_000051", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For error analysis, the 2026-05-08 note says Project Nereid went with Cedar-8B."}, {"sent_id": 3, "text": "The 2026-05-30 compute note links Node Aspen-01 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 4, "text": "Quartz-3B was the model selected by Project Saffron on 2026-06-28 for the claim classification track."}, {"sent_id": 5, "text": "Project Aster had a method update on 2026-04-08."}, {"sent_id": 6, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 7, "text": "Project Anchor logged a personnel update on 2026-05-26: Dr. Arun Kwan took the data steward responsibilities."}, {"sent_id": 8, "text": "On 2026-05-20, Project Anchor shifted Milestone H1: 2026-06-03 became 2026-06-09."}], "source": "synthetic_rewritten", "title": "Project Update 051"}
{"doc_id": "doc_000052", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Pine-07, Quartz-3B produced 0.622 macro F1 on OrionBench for Project Sonata on 2026-04-20."}, {"sent_id": 3, "text": "Project Nereid logged a personnel update on 2026-05-26: Vera Kim took the lead responsibilities."}, {"sent_id": 4, "text": "The 2026-06-10 compute note links Node Willow-05 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 5, "text": "Project Nereid had a method update on 2026-05-13."}, {"sent_id": 6, "text": "The old approach was metric smoothing, and the replacement was hard-negative mining."}, {"sent_id": 7, "text": "The 2026-05-16 compute note links Node Hazel-14 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 8, "text": "Project Sonata received 3 GPUs from Node Maple-01 in the 2026-04-03 allocation log."}], "source": "synthetic_rewritten", "title": "Lab Digest 052"}
{"doc_id": "doc_000053", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Sonata was logged on 2026-05-30."}, {"sent_id": 3, "text": "Node Fir-10 supplied 1 GPU."}, {"sent_id": 4, "text": "The 2026-06-06 compute note links Node Pine-07 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 5, "text": "Project Anchor received 4 GPUs from Node Juniper-06 in the 2026-04-11 allocation log."}, {"sent_id": 6, "text": "Project Saffron had a failed run on 2026-04-03 involving Marble-3B and Node Maple-01."}, {"sent_id": 7, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 8, "text": "Milestone L1 for Project Saffron was revised on 2026-05-20."}, {"sent_id": 9, "text": "The schedule moved from 2026-06-05 to 2026-06-09."}, {"sent_id": 10, "text": "The 2026-05-21 method entry for Project Anchor moved from data mixing to contrastive tuning."}, {"sent_id": 11, "text": "Project Saffron's Milestone X1 deadline change on 2026-06-02 replaced 2026-06-24 with 2026-07-04."}], "source": "synthetic_rewritten", "title": "Project Update 053"}
{"doc_id": "doc_000054", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Marble-8B was the model selected by Project Sonata on 2026-05-24 for the error analysis track."}, {"sent_id": 3, "text": "The 2026-06-26 compute note links Node Poplar-12 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 4, "text": "The 2026-05-09 failure note says Vela-8B on Node Fir-10 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 5, "text": "The staffing note for Project Anchor was dated 2026-06-16."}, {"sent_id": 6, "text": "It named Dr. Arun Kwan for the data steward duties."}, {"sent_id": 7, "text": "Project Saffron ran Aster-7B on SignalSet using Node Willow-05 on 2026-04-04."}, {"sent_id": 8, "text": "The reported accuracy was 0.559."}, {"sent_id": 9, "text": "Cedar-8B was the model selected by Project Meridian on 2026-06-18 for the error analysis track."}], "source": "synthetic_rewritten", "title": "Lab Digest 054"}
{"doc_id": "doc_000055", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata logged a personnel update on 2026-04-27: Samir Kwan took the retrieval owner responsibilities."}, {"sent_id": 3, "text": "The 2026-06-13 method entry for Project Aster moved from chain verification to structured prompting."}, {"sent_id": 4, "text": "The 2026-06-07 method entry for Project Nereid moved from metric smoothing to hard-negative mining."}, {"sent_id": 5, "text": "Project Meridian logged a personnel update on 2026-06-23: Dr. Ravi Reed took the evaluation owner responsibilities."}, {"sent_id": 6, "text": "The 2026-06-04 failure note says Nova-8B on Node Aspen-01 failed for Project Nereid because of an out-of-memory error."}, {"sent_id": 7, "text": "Project Meridian logged a personnel update on 2026-06-22: Mira Sato took the data steward responsibilities."}, {"sent_id": 8, "text": "On 2026-04-01, Project Aster retired QLoRA adaptation in favor of hybrid retrieval."}], "source": "synthetic_rewritten", "title": "Project Update 055"}
{"doc_id": "doc_000056", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Sonata ran Aster-7B on RiverBench-3 using Node Willow-05 on 2026-04-27."}, {"sent_id": 3, "text": "The reported macro F1 was 0.574."}, {"sent_id": 4, "text": "The 2026-04-22 method entry for Project Anchor moved from cross-encoder reranking to late interaction."}, {"sent_id": 5, "text": "The 2026-05-08 compute note links Node Fir-10 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 6, "text": "Project Meridian had a failed run on 2026-05-21 involving Atlas-8B and Node Hazel-14."}, {"sent_id": 7, "text": "The recorded cause was a missing-index error."}, {"sent_id": 8, "text": "Project Aster recorded a model choice for claim classification on 2026-06-05."}, {"sent_id": 9, "text": "The chosen model was Helix-7B."}, {"sent_id": 10, "text": "The 2026-06-23 result for Project Aster put Helix-8B on OrionBench-2 at 0.692 macro F1, with Node Poplar-12 handling the run."}, {"sent_id": 11, "text": "For reranking, the 2026-06-09 note says Project Saffron went with Vela-3B."}, {"sent_id": 12, "text": "The staffing note for Project Aster was dated 2026-06-22."}, {"sent_id": 13, "text": "It named Anika Sato for the lead duties."}], "source": "synthetic_rewritten", "title": "Lab Digest 056"}
{"doc_id": "doc_000057", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Nereid had a method update on 2026-04-01."}, {"sent_id": 3, "text": "The old approach was contrastive tuning, and the replacement was threshold search."}, {"sent_id": 4, "text": "A compute allocation for Project Saffron was logged on 2026-04-30."}, {"sent_id": 5, "text": "Node Maple-01 supplied 2 GPUs."}, {"sent_id": 6, "text": "On 2026-04-12, Project Nereid retired late interaction in favor of BM25 retrieval."}, {"sent_id": 7, "text": "Project Aster recorded a model choice for calibration on 2026-06-28."}, {"sent_id": 8, "text": "The chosen model was Lumen-3B."}, {"sent_id": 9, "text": "Milestone B1 for Project Anchor was revised on 2026-06-10."}, {"sent_id": 10, "text": "The schedule moved from 2026-06-26 to 2026-06-30."}, {"sent_id": 11, "text": "The review note for Pale Compass Study 4 was dated 2026-04-08."}, {"sent_id": 12, "text": "It described the method as data mixing and said it did not use a reward model."}, {"sent_id": 13, "text": "Project Sonata had a failed run on 2026-04-10 involving Quartz-7B and Node Pine-07."}, {"sent_id": 14, "text": "The recorded cause was an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Project Update 057"}
{"doc_id": "doc_000058", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor received 4 GPUs from Node Juniper-06 in the 2026-06-15 allocation log."}, {"sent_id": 3, "text": "Project Anchor logged a personnel update on 2026-05-10: Dr. Arun Bauer took the evaluation owner responsibilities."}, {"sent_id": 4, "text": "The 2026-05-01 failure note says Aster-3B on Node Willow-05 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 5, "text": "Project Aster's Milestone T1 deadline change on 2026-04-15 replaced 2026-05-10 with 2026-05-15."}, {"sent_id": 6, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-13."}, {"sent_id": 7, "text": "The chosen model was Cedar-7B."}, {"sent_id": 8, "text": "Project Saffron recorded a model choice for error analysis on 2026-04-22."}, {"sent_id": 9, "text": "The chosen model was Marble-3B."}, {"sent_id": 10, "text": "The 2026-05-16 result for Project Aster put Lumen-3B on TraceEval at 0.518 evidence F1, with Node Spruce-03 handling the run."}], "source": "synthetic_rewritten", "title": "Lab Digest 058"}
{"doc_id": "doc_000059", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid ran Nova-8B on LabQA using Node Aspen-01 on 2026-06-13."}, {"sent_id": 3, "text": "The reported evidence F1 was 0.690."}, {"sent_id": 4, "text": "Project Saffron recorded a model choice for claim classification on 2026-06-26."}, {"sent_id": 5, "text": "The chosen model was Quartz-7B."}, {"sent_id": 6, "text": "The review note for Pale Compass Study 2 was dated 2026-06-03."}, {"sent_id": 7, "text": "It described the method as contrastive tuning and said it did not use a reward model."}, {"sent_id": 8, "text": "Project Aster recorded a model choice for evidence retrieval on 2026-04-22."}, {"sent_id": 9, "text": "The chosen model was River-7B."}, {"sent_id": 10, "text": "Project Aster ran Lumen-8B on TraceEval using Node Spruce-03 on 2026-05-02."}, {"sent_id": 11, "text": "The reported evidence F1 was 0.848."}, {"sent_id": 12, "text": "On 2026-05-09, Project Meridian retired dense retrieval in favor of rank fusion."}, {"sent_id": 13, "text": "A compute allocation for Project Meridian was logged on 2026-04-20."}, {"sent_id": 14, "text": "Node Laurel-11 supplied 4 GPUs."}, {"sent_id": 15, "text": "Project Anchor had a method update on 2026-05-06."}, {"sent_id": 16, "text": "The old approach was rank fusion, and the replacement was teacher distillation."}], "source": "synthetic_rewritten", "title": "Project Update 059"}
{"doc_id": "doc_000060", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Milestone L1 for Project Saffron was revised on 2026-05-13."}, {"sent_id": 3, "text": "The schedule moved from 2026-06-08 to 2026-06-14."}, {"sent_id": 4, "text": "The 2026-05-30 result for Project Aster put Helix-7B on OrionBench-2 at 0.812 evidence F1, with Node Poplar-12 handling the run."}, {"sent_id": 5, "text": "Project Nereid recorded a model choice for error analysis on 2026-04-26."}, {"sent_id": 6, "text": "The chosen model was Cedar-8B."}, {"sent_id": 7, "text": "A compute allocation for Project Aster was logged on 2026-04-25."}, {"sent_id": 8, "text": "Node Rowan-09 supplied 3 GPUs."}, {"sent_id": 9, "text": "For Project Saffron, the run using Aster-7B on Node Willow-05 did not complete on 2026-06-25; the issue was a missing-index error."}, {"sent_id": 10, "text": "Milestone P1 for Project Meridian was revised on 2026-05-13."}, {"sent_id": 11, "text": "The schedule moved from 2026-05-21 to 2026-05-25."}, {"sent_id": 12, "text": "For Project Nereid, the run using Nova-8B on Node Aspen-01 did not complete on 2026-04-11; the issue was an out-of-memory error."}, {"sent_id": 13, "text": "Project Anchor logged a personnel update on 2026-06-09: Kira Frost took the data steward responsibilities."}, {"sent_id": 14, "text": "The 2026-05-30 method entry for Project Saffron moved from structured prompting to QLoRA adaptation."}], "source": "synthetic_rewritten", "title": "Run Brief 060"}
{"doc_id": "doc_000061", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-05-10 staffing section, Project Saffron listed Lena Costa as the data steward."}, {"sent_id": 3, "text": "Project Aster had a method update on 2026-05-02."}, {"sent_id": 4, "text": "The old approach was LoRA adaptation, and the replacement was alias expansion."}, {"sent_id": 5, "text": "The 2026-06-06 result for Project Saffron put Vela-8B on LumenFacts-3 at 0.595 accuracy, with Node Fir-10 handling the run."}, {"sent_id": 6, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-06-03."}, {"sent_id": 7, "text": "The chosen model was Atlas-3B."}, {"sent_id": 8, "text": "In the 2026-04-07 staffing section, Project Nereid listed Darian Grant as the lead."}, {"sent_id": 9, "text": "Project Meridian received 2 GPUs from Node Laurel-11 in the 2026-04-25 allocation log."}], "source": "synthetic_rewritten", "title": "Review Packet 061"}
{"doc_id": "doc_000062", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid had a failed run on 2026-06-25 involving Nimbus-7B and Node Elm-08."}, {"sent_id": 3, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 4, "text": "For claim classification, the 2026-04-15 note says Project Anchor went with Helix-3B."}, {"sent_id": 5, "text": "The 2026-04-28 result for Project Meridian put Kestrel-8B on CedarQA at 0.777 latency efficiency score, with Node Birch-04 handling the run."}, {"sent_id": 6, "text": "Finch-7B was the model selected by Project Aster on 2026-04-21 for the error analysis track."}, {"sent_id": 7, "text": "Project Nereid's Milestone V1 deadline change on 2026-06-10 replaced 2026-07-01 with 2026-07-06."}, {"sent_id": 8, "text": "On 2026-05-13, Project Saffron retired document chunking in favor of metric smoothing."}, {"sent_id": 9, "text": "In the 2026-04-21 staffing section, Project Nereid listed Darian Grant as the lead."}, {"sent_id": 10, "text": "For calibration, the 2026-05-29 note says Project Saffron went with Mira-3B."}], "source": "synthetic_rewritten", "title": "Run Brief 062"}
{"doc_id": "doc_000063", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Using Node Juniper-06, Finch-8B produced 0.551 accuracy on VestaLogs for Project Anchor on 2026-05-12."}, {"sent_id": 3, "text": "In the 2026-05-05 staffing section, Project Sonata listed Mara Quinn as the lead."}, {"sent_id": 4, "text": "The 2026-05-21 failure note says Cedar-3B on Node Laurel-11 failed for Project Meridian because of a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-04-04 compute note links Node Pine-07 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 6, "text": "Project Saffron ran Mira-3B on MemoTrace-2 using Node Sycamore-13 on 2026-06-02."}, {"sent_id": 7, "text": "The reported accuracy was 0.823."}, {"sent_id": 8, "text": "The 2026-04-22 method entry for Project Saffron moved from LoRA adaptation to alias expansion."}, {"sent_id": 9, "text": "The 2026-06-21 result for Project Sonata put Aster-7B on RiverBench-3 at 0.754 macro F1, with Node Willow-05 handling the run."}, {"sent_id": 10, "text": "The staffing note for Project Saffron was dated 2026-05-25."}, {"sent_id": 11, "text": "It named Dr. Theo Lind for the evaluation owner duties."}, {"sent_id": 12, "text": "Project Aster had a failed run on 2026-06-18 involving Finch-3B and Node Juniper-06."}, {"sent_id": 13, "text": "The recorded cause was an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Review Packet 063"}
{"doc_id": "doc_000064", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "A compute allocation for Project Nereid was logged on 2026-04-18."}, {"sent_id": 3, "text": "Node Laurel-11 supplied 3 GPUs."}, {"sent_id": 4, "text": "Using Node Rowan-09, River-7B produced 0.614 evidence F1 on RiverBench for Project Aster on 2026-04-19."}, {"sent_id": 5, "text": "Project Meridian received 4 GPUs from Node Elm-08 in the 2026-06-06 allocation log."}, {"sent_id": 6, "text": "Project Saffron ran Marble-8B on VestaLogs-3 using Node Maple-01 on 2026-06-22."}, {"sent_id": 7, "text": "The reported latency efficiency score was 0.541."}, {"sent_id": 8, "text": "Project Saffron's Milestone D2 deadline change on 2026-05-29 replaced 2026-06-08 with 2026-06-14."}, {"sent_id": 9, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-04-08."}, {"sent_id": 10, "text": "The chosen model was Aster-7B."}], "source": "synthetic_rewritten", "title": "Run Brief 064"}
{"doc_id": "doc_000065", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Marble-3B was the model selected by Project Saffron on 2026-04-24 for the error analysis track."}, {"sent_id": 3, "text": "Project Saffron had a failed run on 2026-05-14 involving Quartz-8B and Node Pine-07."}, {"sent_id": 4, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-05-24 method entry for Project Meridian moved from hybrid retrieval to document chunking."}, {"sent_id": 6, "text": "The 2026-04-04 compute note links Node Birch-04 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 7, "text": "On 2026-04-08, Project Aster shifted Milestone H1: 2026-04-17 became 2026-04-22."}, {"sent_id": 8, "text": "The 2026-05-31 method entry for Project Aster moved from alias expansion to dense retrieval."}, {"sent_id": 9, "text": "The review note for Nested Verifier Study 2 was dated 2026-06-19."}, {"sent_id": 10, "text": "It described the method as calibrated voting and said it did not use a reward model."}, {"sent_id": 11, "text": "Project Anchor ran Finch-7B on VestaLogs using Node Juniper-06 on 2026-06-06."}, {"sent_id": 12, "text": "The reported latency efficiency score was 0.881."}, {"sent_id": 13, "text": "Project Saffron received 4 GPUs from Node Willow-05 in the 2026-06-13 allocation log."}], "source": "synthetic_rewritten", "title": "Review Packet 065"}
{"doc_id": "doc_000066", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For Project Anchor, the run using River-3B on Node Rowan-09 did not complete on 2026-04-24; the issue was a missing-index error."}, {"sent_id": 3, "text": "Project Anchor logged a personnel update on 2026-05-03: Dr. Arun Kwan took the data steward responsibilities."}, {"sent_id": 4, "text": "Quartz-8B was the model selected by Project Saffron on 2026-06-17 for the claim classification track."}, {"sent_id": 5, "text": "Project Sonata's Milestone R1 deadline change on 2026-04-10 replaced 2026-04-19 with 2026-04-24."}, {"sent_id": 6, "text": "The review note for Stable Chains Study 1 was dated 2026-06-17."}, {"sent_id": 7, "text": "It described the method as document chunking and said it did not use a reward model."}, {"sent_id": 8, "text": "In the 2026-05-05 staffing section, Project Meridian listed Nadia Chen as the evaluation owner."}, {"sent_id": 9, "text": "The staffing note for Project Aster was dated 2026-05-19."}, {"sent_id": 10, "text": "It named Anika Costa for the retrieval owner duties."}, {"sent_id": 11, "text": "Project Anchor ran Finch-3B on VestaLogs using Node Juniper-06 on 2026-04-20."}, {"sent_id": 12, "text": "The reported accuracy was 0.791."}], "source": "synthetic_rewritten", "title": "Run Brief 066"}
{"doc_id": "doc_000067", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian ran Nova-8B on LumenFacts-2 using Node Aspen-01 on 2026-04-21."}, {"sent_id": 3, "text": "The reported latency efficiency score was 0.825."}, {"sent_id": 4, "text": "Kestrel-8B was the model selected by Project Nereid on 2026-04-12 for the claim classification track."}, {"sent_id": 5, "text": "Nimbus-3B was the model selected by Project Nereid on 2026-04-26 for the calibration track."}, {"sent_id": 6, "text": "Pale Compass Study 1 appeared in a 2026-04-03 review entry using data mixing; the note said it did not use a reward model."}, {"sent_id": 7, "text": "For evidence retrieval, the 2026-06-24 note says Project Sonata went with Aster-7B."}, {"sent_id": 8, "text": "Project Meridian recorded a model choice for error analysis on 2026-06-26."}, {"sent_id": 9, "text": "The chosen model was Cedar-3B."}, {"sent_id": 10, "text": "Project Anchor ran Helix-3B on CedarQA-3 using Node Poplar-12 on 2026-06-29."}, {"sent_id": 11, "text": "The reported latency efficiency score was 0.857."}], "source": "synthetic_rewritten", "title": "Review Packet 067"}
{"doc_id": "doc_000068", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-24 compute note links Node Fir-10 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "For Project Aster, the run using Orchid-7B on Node Cedar-02 did not complete on 2026-05-01; the issue was an unstable-validation-loss error."}, {"sent_id": 4, "text": "The 2026-06-26 failure note says Helix-3B on Node Poplar-12 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-04-17 compute note links Node Rowan-09 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 6, "text": "Using Node Laurel-11, Cedar-8B produced 0.681 accuracy on VestaLogs-2 for Project Meridian on 2026-05-25."}, {"sent_id": 7, "text": "Using Node Cedar-02, Orchid-8B produced 0.890 evidence F1 on LabQA-3 for Project Aster on 2026-04-21."}, {"sent_id": 8, "text": "Project Anchor had a method update on 2026-05-30."}, {"sent_id": 9, "text": "The old approach was QLoRA adaptation, and the replacement was hybrid retrieval."}, {"sent_id": 10, "text": "Project Meridian recorded a model choice for reranking on 2026-04-30."}, {"sent_id": 11, "text": "The chosen model was Nova-8B."}, {"sent_id": 12, "text": "Nimbus-3B was the model selected by Project Meridian on 2026-06-03 for the calibration track."}], "source": "synthetic_rewritten", "title": "Run Brief 068"}
{"doc_id": "doc_000069", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian had a failed run on 2026-05-01 involving Kestrel-8B and Node Birch-04."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "The 2026-04-04 result for Project Anchor put Orchid-8B on LumenFacts at 0.845 latency efficiency score, with Node Cedar-02 handling the run."}, {"sent_id": 5, "text": "Mira-8B was the model selected by Project Sonata on 2026-04-29 for the calibration track."}, {"sent_id": 6, "text": "Project Aster ran Lumen-7B on TraceEval using Node Spruce-03 on 2026-06-30."}, {"sent_id": 7, "text": "The reported evidence F1 was 0.878."}, {"sent_id": 8, "text": "Milestone T1 for Project Anchor was revised on 2026-05-26."}, {"sent_id": 9, "text": "The schedule moved from 2026-06-05 to 2026-06-11."}, {"sent_id": 10, "text": "The 2026-06-11 failure note says Cedar-3B on Node Laurel-11 failed for Project Meridian because of a missing-index error."}, {"sent_id": 11, "text": "For Project Sonata, the run using Vela-7B on Node Fir-10 did not complete on 2026-06-12; the issue was an unstable-validation-loss error."}, {"sent_id": 12, "text": "The 2026-04-10 failure note says Orchid-7B on Node Cedar-02 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 13, "text": "In the 2026-05-25 staffing section, Project Saffron listed Owen Torres as the evaluation owner."}], "source": "synthetic_rewritten", "title": "Review Packet 069"}
{"doc_id": "doc_000070", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Aster had a method update on 2026-05-20."}, {"sent_id": 3, "text": "The old approach was contrastive tuning, and the replacement was threshold search."}, {"sent_id": 4, "text": "Project Sonata had a method update on 2026-06-20."}, {"sent_id": 5, "text": "The old approach was QLoRA adaptation, and the replacement was hybrid retrieval."}, {"sent_id": 6, "text": "River-7B was the model selected by Project Anchor on 2026-05-06 for the evidence retrieval track."}, {"sent_id": 7, "text": "Project Nereid had a failed run on 2026-05-02 involving Kestrel-3B and Node Birch-04."}, {"sent_id": 8, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 9, "text": "Project Aster had a method update on 2026-05-07."}, {"sent_id": 10, "text": "The old approach was LoRA adaptation, and the replacement was alias expansion."}, {"sent_id": 11, "text": "Project Sonata logged a personnel update on 2026-05-04: Samir Ames took the lead responsibilities."}], "source": "synthetic_rewritten", "title": "Project Update 070"}
{"doc_id": "doc_000071", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Orchid-3B was the model selected by Project Anchor on 2026-06-21 for the reranking track."}, {"sent_id": 3, "text": "On 2026-06-19, Project Meridian shifted Milestone P1: 2026-06-27 became 2026-07-05."}, {"sent_id": 4, "text": "Milestone D2 for Project Sonata was revised on 2026-06-03."}, {"sent_id": 5, "text": "The schedule moved from 2026-06-18 to 2026-06-14."}, {"sent_id": 6, "text": "For Project Anchor, the run using Orchid-3B on Node Cedar-02 did not complete on 2026-06-18; the issue was a missing-index error."}, {"sent_id": 7, "text": "The 2026-06-23 result for Project Nereid put Nimbus-3B on TraceEval-2 at 0.588 evidence F1, with Node Elm-08 handling the run."}, {"sent_id": 8, "text": "In the 2026-05-19 staffing section, Project Saffron listed Lena Costa as the data steward."}], "source": "synthetic_rewritten", "title": "Lab Digest 071"}
{"doc_id": "doc_000072", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For error analysis, the 2026-05-29 note says Project Anchor went with Finch-3B."}, {"sent_id": 3, "text": "Project Aster had a failed run on 2026-05-28 involving River-7B and Node Rowan-09."}, {"sent_id": 4, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 5, "text": "Milestone T1 for Project Aster was revised on 2026-05-15."}, {"sent_id": 6, "text": "The schedule moved from 2026-05-30 to 2026-05-26."}, {"sent_id": 7, "text": "Orchid-3B was the model selected by Project Aster on 2026-05-08 for the reranking track."}, {"sent_id": 8, "text": "Project Nereid ran Cedar-7B on NereidNotes-3 using Node Laurel-11 on 2026-04-21."}, {"sent_id": 9, "text": "The reported macro F1 was 0.786."}, {"sent_id": 10, "text": "Project Anchor recorded a model choice for claim classification on 2026-05-08."}, {"sent_id": 11, "text": "The chosen model was Helix-3B."}, {"sent_id": 12, "text": "Project Saffron had a method update on 2026-05-28."}, {"sent_id": 13, "text": "The old approach was LoRA adaptation, and the replacement was alias expansion."}, {"sent_id": 14, "text": "Helix-8B was the model selected by Project Anchor on 2026-04-24 for the claim classification track."}], "source": "synthetic_rewritten", "title": "Project Update 072"}
{"doc_id": "doc_000073", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-12, Project Sonata retired rank fusion in favor of teacher distillation."}, {"sent_id": 3, "text": "Project Aster logged a personnel update on 2026-05-03: Mina Adler took the retrieval owner responsibilities."}, {"sent_id": 4, "text": "On 2026-06-10, Project Aster shifted Milestone Z1: 2026-07-03 became 2026-07-01."}, {"sent_id": 5, "text": "Vela-3B was the model selected by Project Sonata on 2026-05-01 for the reranking track."}, {"sent_id": 6, "text": "Project Meridian recorded a model choice for calibration on 2026-06-28."}, {"sent_id": 7, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 8, "text": "Vela-8B was the model selected by Project Saffron on 2026-04-01 for the reranking track."}], "source": "synthetic_rewritten", "title": "Lab Digest 073"}
{"doc_id": "doc_000074", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-04-23 failure note says Mira-8B on Node Sycamore-13 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 3, "text": "A compute allocation for Project Anchor was logged on 2026-05-02."}, {"sent_id": 4, "text": "Node Juniper-06 supplied 2 GPUs."}, {"sent_id": 5, "text": "Project Nereid received 1 GPU from Node Aspen-01 in the 2026-06-19 allocation log."}, {"sent_id": 6, "text": "The 2026-05-31 method entry for Project Anchor moved from rank fusion to teacher distillation."}, {"sent_id": 7, "text": "On 2026-04-26, Project Anchor retired teacher distillation in favor of query rewriting."}, {"sent_id": 8, "text": "The review note for Amber Ranking Study 3 was dated 2026-06-05."}, {"sent_id": 9, "text": "It described the method as teacher distillation and said it did not use a reward model."}, {"sent_id": 10, "text": "Project Anchor received 4 GPUs from Node Spruce-03 in the 2026-04-18 allocation log."}], "source": "synthetic_rewritten", "title": "Project Update 074"}
{"doc_id": "doc_000075", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron recorded a model choice for error analysis on 2026-05-03."}, {"sent_id": 3, "text": "The chosen model was Marble-7B."}, {"sent_id": 4, "text": "Project Nereid had a method update on 2026-04-12."}, {"sent_id": 5, "text": "The old approach was alias expansion, and the replacement was dense retrieval."}, {"sent_id": 6, "text": "The 2026-05-23 result for Project Nereid put Cedar-8B on NereidNotes-3 at 0.756 macro F1, with Node Laurel-11 handling the run."}, {"sent_id": 7, "text": "Project Nereid received 3 GPUs from Node Birch-04 in the 2026-04-29 allocation log."}, {"sent_id": 8, "text": "The 2026-05-31 method entry for Project Aster moved from late interaction to BM25 retrieval."}, {"sent_id": 9, "text": "Marble-8B was the model selected by Project Saffron on 2026-05-07 for the error analysis track."}, {"sent_id": 10, "text": "Project Aster had a method update on 2026-05-09."}, {"sent_id": 11, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 12, "text": "Project Saffron recorded a model choice for reranking on 2026-06-07."}, {"sent_id": 13, "text": "The chosen model was Vela-8B."}, {"sent_id": 14, "text": "The 2026-06-08 result for Project Saffron put Quartz-7B on CedarQA-2 at 0.637 latency efficiency score, with Node Pine-07 handling the run."}], "source": "synthetic_rewritten", "title": "Lab Digest 075"}
{"doc_id": "doc_000076", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for calibration on 2026-05-21."}, {"sent_id": 3, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 4, "text": "Project Sonata ran Quartz-8B on OrionBench using Node Pine-07 on 2026-04-21."}, {"sent_id": 5, "text": "The reported evidence F1 was 0.682."}, {"sent_id": 6, "text": "Project Meridian had a method update on 2026-05-14."}, {"sent_id": 7, "text": "The old approach was dense retrieval, and the replacement was rank fusion."}, {"sent_id": 8, "text": "Project Saffron recorded a model choice for reranking on 2026-04-15."}, {"sent_id": 9, "text": "The chosen model was Vela-3B."}, {"sent_id": 10, "text": "Project Saffron's Milestone F1 deadline change on 2026-05-01 replaced 2026-05-19 with 2026-05-29."}, {"sent_id": 11, "text": "In the 2026-04-06 staffing section, Project Saffron listed Dr. Theo Lind as the evaluation owner."}], "source": "synthetic_rewritten", "title": "Project Update 076"}
{"doc_id": "doc_000077", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Using Node Rowan-09, River-7B produced 0.809 latency efficiency score on SignalSet-2 for Project Anchor on 2026-05-02."}, {"sent_id": 3, "text": "On 2026-05-15, Project Meridian shifted Milestone J1: 2026-06-04 became 2026-06-08."}, {"sent_id": 4, "text": "For Project Nereid, the run using Nimbus-8B on Node Elm-08 did not complete on 2026-06-27; the issue was an unstable-validation-loss error."}, {"sent_id": 5, "text": "Project Nereid had a failed run on 2026-04-25 involving Kestrel-7B and Node Birch-04."}, {"sent_id": 6, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 7, "text": "Project Nereid's Milestone B2 deadline change on 2026-04-27 replaced 2026-05-16 with 2026-05-13."}, {"sent_id": 8, "text": "Milestone D2 for Project Sonata was revised on 2026-06-10."}, {"sent_id": 9, "text": "The schedule moved from 2026-07-05 to 2026-07-14."}], "source": "synthetic_rewritten", "title": "Lab Digest 077"}
{"doc_id": "doc_000078", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Aspen-01, Nova-3B produced 0.675 latency efficiency score on LumenFacts-2 for Project Meridian on 2026-06-29."}, {"sent_id": 3, "text": "On 2026-05-13, Project Aster retired chain verification in favor of structured prompting."}, {"sent_id": 4, "text": "The staffing note for Project Nereid was dated 2026-05-12."}, {"sent_id": 5, "text": "It named Nora Sol for the retrieval owner duties."}, {"sent_id": 6, "text": "The staffing note for Project Anchor was dated 2026-06-30."}, {"sent_id": 7, "text": "It named Kira Iyer for the evaluation owner duties."}, {"sent_id": 8, "text": "For claim classification, the 2026-04-03 note says Project Saffron went with Quartz-3B."}, {"sent_id": 9, "text": "Project Aster had a method update on 2026-04-25."}, {"sent_id": 10, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 11, "text": "Vela-8B was the model selected by Project Saffron on 2026-05-10 for the reranking track."}], "source": "synthetic_rewritten", "title": "Project Update 078"}
{"doc_id": "doc_000079", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-27, the Nested Verifier Study 1 review reported QLoRA adaptation and said it did not use a reward model."}, {"sent_id": 3, "text": "The 2026-04-06 result for Project Saffron put Mira-8B on MemoTrace-2 at 0.523 accuracy, with Node Sycamore-13 handling the run."}, {"sent_id": 4, "text": "The 2026-05-24 method entry for Project Aster moved from rank fusion to teacher distillation."}, {"sent_id": 5, "text": "Milestone T1 for Project Anchor was revised on 2026-05-20."}, {"sent_id": 6, "text": "The schedule moved from 2026-06-09 to 2026-06-17."}, {"sent_id": 7, "text": "The 2026-05-24 method entry for Project Aster moved from dense retrieval to rank fusion."}, {"sent_id": 8, "text": "Project Anchor logged a personnel update on 2026-05-18: Selene Kim took the evaluation owner responsibilities."}, {"sent_id": 9, "text": "On 2026-06-11, Project Saffron retired evidence pooling in favor of calibrated voting."}], "source": "synthetic_rewritten", "title": "Lab Digest 079"}
{"doc_id": "doc_000080", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Anchor recorded a model choice for error analysis on 2026-05-10."}, {"sent_id": 3, "text": "The chosen model was Finch-7B."}, {"sent_id": 4, "text": "Project Anchor ran Lumen-3B on MemoTrace-3 using Node Spruce-03 on 2026-06-20."}, {"sent_id": 5, "text": "The reported accuracy was 0.863."}, {"sent_id": 6, "text": "Using Node Sycamore-13, Mira-3B produced 0.718 macro F1 on TraceEval-3 for Project Sonata on 2026-04-19."}, {"sent_id": 7, "text": "Using Node Juniper-06, Finch-3B produced 0.881 latency efficiency score on VestaLogs for Project Anchor on 2026-05-11."}, {"sent_id": 8, "text": "Project Sonata had a method update on 2026-06-21."}, {"sent_id": 9, "text": "The old approach was rank fusion, and the replacement was teacher distillation."}, {"sent_id": 10, "text": "Project Sonata had a method update on 2026-05-24."}, {"sent_id": 11, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 12, "text": "The 2026-05-18 compute note links Node Spruce-03 to Project Aster with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Review Packet 080"}
{"doc_id": "doc_000081", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian had a failed run on 2026-06-25 involving Nimbus-8B and Node Elm-08."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "The 2026-05-02 compute note links Node Pine-07 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 5, "text": "The 2026-05-03 result for Project Aster put Finch-7B on NereidNotes-2 at 0.596 macro F1, with Node Juniper-06 handling the run."}, {"sent_id": 6, "text": "Nimbus-8B was the model selected by Project Nereid on 2026-06-11 for the calibration track."}, {"sent_id": 7, "text": "Cedar-3B was the model selected by Project Nereid on 2026-04-19 for the error analysis track."}, {"sent_id": 8, "text": "The 2026-04-25 compute note links Node Birch-04 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 9, "text": "Project Anchor ran River-7B on SignalSet-2 using Node Rowan-09 on 2026-04-14."}, {"sent_id": 10, "text": "The reported latency efficiency score was 0.899."}], "source": "synthetic_rewritten", "title": "Run Brief 081"}
{"doc_id": "doc_000082", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For error analysis, the 2026-04-01 note says Project Meridian went with Cedar-8B."}, {"sent_id": 3, "text": "The 2026-06-08 result for Project Aster put Finch-7B on NereidNotes-2 at 0.806 macro F1, with Node Juniper-06 handling the run."}, {"sent_id": 4, "text": "On 2026-05-06, Project Aster retired alias expansion in favor of dense retrieval."}, {"sent_id": 5, "text": "On 2026-06-13, Project Saffron retired sentence pruning in favor of cross-encoder reranking."}, {"sent_id": 6, "text": "A compute allocation for Project Aster was logged on 2026-05-01."}, {"sent_id": 7, "text": "Node Juniper-06 supplied 1 GPU."}, {"sent_id": 8, "text": "The 2026-04-17 compute note links Node Fir-10 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 9, "text": "Helix-3B was the model selected by Project Anchor on 2026-04-23 for the claim classification track."}, {"sent_id": 10, "text": "Nova-7B was the model selected by Project Meridian on 2026-05-01 for the reranking track."}, {"sent_id": 11, "text": "Project Saffron had a method update on 2026-06-07."}, {"sent_id": 12, "text": "The old approach was sentence pruning, and the replacement was cross-encoder reranking."}], "source": "synthetic_rewritten", "title": "Review Packet 082"}
{"doc_id": "doc_000083", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-21."}, {"sent_id": 3, "text": "The chosen model was Cedar-7B."}, {"sent_id": 4, "text": "Project Anchor recorded a model choice for reranking on 2026-04-03."}, {"sent_id": 5, "text": "The chosen model was Orchid-3B."}, {"sent_id": 6, "text": "The 2026-04-19 method entry for Project Aster moved from cross-encoder reranking to late interaction."}, {"sent_id": 7, "text": "Project Saffron had a failed run on 2026-06-12 involving Aster-8B and Node Willow-05."}, {"sent_id": 8, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "The 2026-06-15 result for Project Sonata put Vela-7B on LabQA-2 at 0.550 macro F1, with Node Fir-10 handling the run."}, {"sent_id": 10, "text": "On 2026-06-17, Project Anchor retired query rewriting in favor of sentence pruning."}, {"sent_id": 11, "text": "Milestone N1 for Project Aster was revised on 2026-06-25."}, {"sent_id": 12, "text": "The schedule moved from 2026-07-12 to 2026-07-21."}], "source": "synthetic_rewritten", "title": "Run Brief 083"}
{"doc_id": "doc_000084", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-24 method entry for Project Anchor moved from alias expansion to dense retrieval."}, {"sent_id": 3, "text": "Cedar-3B was the model selected by Project Nereid on 2026-05-07 for the error analysis track."}, {"sent_id": 4, "text": "On 2026-04-15, Project Nereid shifted Milestone B2: 2026-04-24 became 2026-05-03."}, {"sent_id": 5, "text": "The 2026-06-24 method entry for Project Saffron moved from threshold search to reward reranking."}, {"sent_id": 6, "text": "Project Nereid had a method update on 2026-05-03."}, {"sent_id": 7, "text": "The old approach was reward reranking, and the replacement was data mixing."}, {"sent_id": 8, "text": "Using Node Hazel-14, Atlas-8B produced 0.684 macro F1 on RiverBench-2 for Project Nereid on 2026-06-22."}, {"sent_id": 9, "text": "In the 2026-04-19 staffing section, Project Meridian listed Camila Brooks as the data steward."}, {"sent_id": 10, "text": "Project Meridian recorded a model choice for claim classification on 2026-05-20."}, {"sent_id": 11, "text": "The chosen model was Kestrel-8B."}, {"sent_id": 12, "text": "The review note for Nested Verifier Study 1 was dated 2026-05-22."}, {"sent_id": 13, "text": "It described the method as cross-encoder reranking and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Review Packet 084"}
{"doc_id": "doc_000085", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Elm-08, Nimbus-8B produced 0.768 macro F1 on TraceEval-2 for Project Nereid on 2026-05-18."}, {"sent_id": 3, "text": "The 2026-04-26 result for Project Anchor put Orchid-8B on LumenFacts at 0.605 accuracy, with Node Cedar-02 handling the run."}, {"sent_id": 4, "text": "Project Aster had a failed run on 2026-04-24 involving Orchid-8B and Node Cedar-02."}, {"sent_id": 5, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 6, "text": "For claim classification, the 2026-04-21 note says Project Nereid went with Kestrel-8B."}, {"sent_id": 7, "text": "Project Sonata had a failed run on 2026-04-23 involving Aster-8B and Node Willow-05."}, {"sent_id": 8, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 9, "text": "Project Sonata received 3 GPUs from Node Sycamore-13 in the 2026-06-13 allocation log."}, {"sent_id": 10, "text": "The 2026-04-04 compute note links Node Poplar-12 to Project Aster with an allocation of 3 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 085"}
{"doc_id": "doc_000086", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For evidence retrieval, the 2026-04-19 note says Project Anchor went with River-8B."}, {"sent_id": 3, "text": "In the 2026-06-16 staffing section, Project Saffron listed Petra Adler as the data steward."}, {"sent_id": 4, "text": "The staffing note for Project Sonata was dated 2026-04-19."}, {"sent_id": 5, "text": "It named Mara Lane for the retrieval owner duties."}, {"sent_id": 6, "text": "On 2026-05-06, Project Anchor shifted Milestone Z1: 2026-05-14 became 2026-05-22."}, {"sent_id": 7, "text": "Project Sonata had a failed run on 2026-06-11 involving Vela-8B and Node Fir-10."}, {"sent_id": 8, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 9, "text": "The 2026-06-23 result for Project Saffron put Aster-3B on SignalSet at 0.679 latency efficiency score, with Node Willow-05 handling the run."}, {"sent_id": 10, "text": "For Project Anchor, the run using Lumen-3B on Node Spruce-03 did not complete on 2026-05-21; the issue was a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Review Packet 086"}
{"doc_id": "doc_000087", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian received 2 GPUs from Node Elm-08 in the 2026-05-06 allocation log."}, {"sent_id": 3, "text": "The 2026-05-03 method entry for Project Nereid moved from QLoRA adaptation to hybrid retrieval."}, {"sent_id": 4, "text": "On 2026-05-31, Project Sonata retired sentence pruning in favor of cross-encoder reranking."}, {"sent_id": 5, "text": "For evidence retrieval, the 2026-06-26 note says Project Meridian went with Atlas-8B."}, {"sent_id": 6, "text": "Milestone J1 for Project Nereid was revised on 2026-05-12."}, {"sent_id": 7, "text": "The schedule moved from 2026-05-27 to 2026-05-23."}, {"sent_id": 8, "text": "Project Aster had a method update on 2026-04-12."}, {"sent_id": 9, "text": "The old approach was document chunking, and the replacement was metric smoothing."}], "source": "synthetic_rewritten", "title": "Run Brief 087"}
{"doc_id": "doc_000088", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Lumen-8B was the model selected by Project Anchor on 2026-06-10 for the calibration track."}, {"sent_id": 3, "text": "For evidence retrieval, the 2026-06-10 note says Project Nereid went with Atlas-8B."}, {"sent_id": 4, "text": "The 2026-06-30 result for Project Meridian put Nova-8B on LumenFacts-2 at 0.735 accuracy, with Node Aspen-01 handling the run."}, {"sent_id": 5, "text": "Project Nereid had a failed run on 2026-04-30 involving Nimbus-8B and Node Elm-08."}, {"sent_id": 6, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 7, "text": "On 2026-04-20, Project Nereid shifted Milestone B2: 2026-04-29 became 2026-05-04."}, {"sent_id": 8, "text": "The 2026-06-21 result for Project Nereid put Nimbus-8B on TraceEval-2 at 0.858 evidence F1, with Node Elm-08 handling the run."}, {"sent_id": 9, "text": "Milestone D1 for Project Nereid was revised on 2026-06-22."}, {"sent_id": 10, "text": "The schedule moved from 2026-07-09 to 2026-07-18."}, {"sent_id": 11, "text": "The 2026-04-28 result for Project Nereid put Kestrel-7B on OrionBench-3 at 0.582 macro F1, with Node Birch-04 handling the run."}], "source": "synthetic_rewritten", "title": "Review Packet 088"}
{"doc_id": "doc_000089", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-15, Project Sonata shifted Milestone L1: 2026-05-06 became 2026-05-15."}, {"sent_id": 3, "text": "Using Node Pine-07, Quartz-3B produced 0.817 latency efficiency score on CedarQA-2 for Project Saffron on 2026-04-07."}, {"sent_id": 4, "text": "Project Sonata recorded a model choice for evidence retrieval on 2026-05-15."}, {"sent_id": 5, "text": "The chosen model was Aster-7B."}, {"sent_id": 6, "text": "The 2026-05-23 compute note links Node Rowan-09 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 7, "text": "The 2026-06-09 result for Project Aster put Orchid-8B on LabQA-3 at 0.710 evidence F1, with Node Cedar-02 handling the run."}, {"sent_id": 8, "text": "Project Anchor logged a personnel update on 2026-06-02: Iris Lane took the evaluation owner responsibilities."}, {"sent_id": 9, "text": "Milestone Z1 for Project Aster was revised on 2026-04-16."}, {"sent_id": 10, "text": "The schedule moved from 2026-04-29 to 2026-05-04."}, {"sent_id": 11, "text": "The 2026-06-19 failure note says Aster-7B on Node Willow-05 failed for Project Saffron because of a missing-index error."}, {"sent_id": 12, "text": "Using Node Sycamore-13, Mira-8B produced 0.748 evidence F1 on TraceEval-3 for Project Sonata on 2026-04-26."}], "source": "synthetic_rewritten", "title": "Run Brief 089"}
{"doc_id": "doc_000090", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian recorded a model choice for reranking on 2026-06-17."}, {"sent_id": 3, "text": "The chosen model was Nova-3B."}, {"sent_id": 4, "text": "The 2026-05-28 failure note says Quartz-3B on Node Pine-07 failed for Project Saffron because of a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-06-07 method entry for Project Anchor moved from contrastive tuning to threshold search."}, {"sent_id": 6, "text": "Project Nereid had a method update on 2026-05-10."}, {"sent_id": 7, "text": "The old approach was rank fusion, and the replacement was teacher distillation."}, {"sent_id": 8, "text": "The 2026-06-11 method entry for Project Aster moved from alias expansion to dense retrieval."}, {"sent_id": 9, "text": "Milestone D1 for Project Meridian was revised on 2026-06-19."}, {"sent_id": 10, "text": "The schedule moved from 2026-07-11 to 2026-07-17."}, {"sent_id": 11, "text": "Project Anchor had a failed run on 2026-04-23 involving Orchid-8B and Node Cedar-02."}, {"sent_id": 12, "text": "The recorded cause was a missing-index error."}], "source": "synthetic_rewritten", "title": "Lab Digest 090"}
{"doc_id": "doc_000091", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Using Node Elm-08, Nimbus-3B produced 0.873 accuracy on MemoTrace for Project Meridian on 2026-05-10."}, {"sent_id": 3, "text": "Project Nereid ran Nova-3B on LabQA using Node Aspen-01 on 2026-05-19."}, {"sent_id": 4, "text": "The reported macro F1 was 0.750."}, {"sent_id": 5, "text": "The staffing note for Project Meridian was dated 2026-06-16."}, {"sent_id": 6, "text": "It named Elian Ames for the evaluation owner duties."}, {"sent_id": 7, "text": "Project Nereid had a method update on 2026-04-18."}, {"sent_id": 8, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 9, "text": "In the 2026-05-26 staffing section, Project Nereid listed Dr. Julian Stone as the retrieval owner."}, {"sent_id": 10, "text": "Milestone F1 for Project Saffron was revised on 2026-04-24."}, {"sent_id": 11, "text": "The schedule moved from 2026-05-02 to 2026-05-06."}, {"sent_id": 12, "text": "For Project Aster, the run using Finch-7B on Node Juniper-06 did not complete on 2026-06-26; the issue was an out-of-memory error."}, {"sent_id": 13, "text": "The staffing note for Project Aster was dated 2026-04-13."}, {"sent_id": 14, "text": "It named Felix Brooks for the retrieval owner duties."}], "source": "synthetic_rewritten", "title": "Project Update 091"}
{"doc_id": "doc_000092", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Kestrel-3B was the model selected by Project Nereid on 2026-05-20 for the claim classification track."}, {"sent_id": 3, "text": "Project Sonata recorded a model choice for calibration on 2026-04-07."}, {"sent_id": 4, "text": "The chosen model was Mira-3B."}, {"sent_id": 5, "text": "The staffing note for Project Anchor was dated 2026-05-04."}, {"sent_id": 6, "text": "It named Leo Park for the evaluation owner duties."}, {"sent_id": 7, "text": "Project Anchor had a failed run on 2026-05-15 involving Finch-8B and Node Juniper-06."}, {"sent_id": 8, "text": "The recorded cause was a missing-index error."}, {"sent_id": 9, "text": "The 2026-05-02 method entry for Project Anchor moved from BM25 retrieval to evidence pooling."}, {"sent_id": 10, "text": "Project Aster recorded a model choice for reranking on 2026-05-28."}, {"sent_id": 11, "text": "The chosen model was Orchid-7B."}], "source": "synthetic_rewritten", "title": "Lab Digest 092"}
{"doc_id": "doc_000093", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Aster recorded a model choice for error analysis on 2026-04-15."}, {"sent_id": 3, "text": "The chosen model was Finch-7B."}, {"sent_id": 4, "text": "The review note for Pale Compass Study 3 was dated 2026-05-29."}, {"sent_id": 5, "text": "It described the method as contrastive tuning and said it did not use a reward model."}, {"sent_id": 6, "text": "For claim classification, the 2026-04-29 note says Project Sonata went with Quartz-7B."}, {"sent_id": 7, "text": "Milestone F1 for Project Saffron was revised on 2026-04-30."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-08 to 2026-05-16."}, {"sent_id": 9, "text": "Project Nereid recorded a model choice for calibration on 2026-05-28."}, {"sent_id": 10, "text": "The chosen model was Nimbus-7B."}, {"sent_id": 11, "text": "Using Node Poplar-12, Helix-8B produced 0.842 macro F1 on OrionBench-2 for Project Aster on 2026-05-11."}, {"sent_id": 12, "text": "Project Nereid logged a personnel update on 2026-04-06: Darian Hale took the retrieval owner responsibilities."}, {"sent_id": 13, "text": "Milestone D2 for Project Sonata was revised on 2026-04-23."}, {"sent_id": 14, "text": "The schedule moved from 2026-05-18 to 2026-05-27."}], "source": "synthetic_rewritten", "title": "Project Update 093"}
{"doc_id": "doc_000094", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Finch-7B was the model selected by Project Anchor on 2026-06-12 for the error analysis track."}, {"sent_id": 3, "text": "The 2026-04-23 failure note says Mira-3B on Node Sycamore-13 failed for Project Saffron because of a missing-index error."}, {"sent_id": 4, "text": "For Project Nereid, the run using Nimbus-8B on Node Elm-08 did not complete on 2026-04-24; the issue was an out-of-memory error."}, {"sent_id": 5, "text": "Project Sonata recorded a model choice for error analysis on 2026-06-10."}, {"sent_id": 6, "text": "The chosen model was Marble-7B."}, {"sent_id": 7, "text": "Using Node Rowan-09, River-3B produced 0.839 latency efficiency score on SignalSet-2 for Project Anchor on 2026-06-30."}, {"sent_id": 8, "text": "Project Aster logged a personnel update on 2026-05-18: Felix Brooks took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 094"}
{"doc_id": "doc_000095", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for error analysis on 2026-06-24."}, {"sent_id": 3, "text": "The chosen model was Cedar-7B."}, {"sent_id": 4, "text": "The 2026-04-16 compute note links Node Poplar-12 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 5, "text": "Project Meridian logged a personnel update on 2026-04-20: Dr. Ravi Reed took the evaluation owner responsibilities."}, {"sent_id": 6, "text": "A compute allocation for Project Aster was logged on 2026-04-18."}, {"sent_id": 7, "text": "Node Poplar-12 supplied 3 GPUs."}, {"sent_id": 8, "text": "Project Sonata recorded a model choice for calibration on 2026-04-12."}, {"sent_id": 9, "text": "The chosen model was Mira-7B."}, {"sent_id": 10, "text": "The 2026-04-07 result for Project Anchor put Lumen-8B on MemoTrace-3 at 0.713 accuracy, with Node Spruce-03 handling the run."}, {"sent_id": 11, "text": "The staffing note for Project Saffron was dated 2026-04-06."}, {"sent_id": 12, "text": "It named Priya Moss for the data steward duties."}, {"sent_id": 13, "text": "Project Sonata logged a personnel update on 2026-06-22: Sofia Nadir took the retrieval owner responsibilities."}, {"sent_id": 14, "text": "Vela-3B was the model selected by Project Sonata on 2026-06-28 for the reranking track."}], "source": "synthetic_rewritten", "title": "Project Update 095"}
{"doc_id": "doc_000096", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-04-07 staffing section, Project Meridian listed Camila Quinn as the evaluation owner."}, {"sent_id": 3, "text": "On 2026-04-26, Project Nereid retired contrastive tuning in favor of threshold search."}, {"sent_id": 4, "text": "Project Nereid received 3 GPUs from Node Elm-08 in the 2026-06-17 allocation log."}, {"sent_id": 5, "text": "On 2026-04-22, Project Meridian shifted Milestone V1: 2026-05-18 became 2026-05-24."}, {"sent_id": 6, "text": "Project Aster ran Finch-8B on NereidNotes-2 using Node Juniper-06 on 2026-04-25."}, {"sent_id": 7, "text": "The reported evidence F1 was 0.896."}, {"sent_id": 8, "text": "A compute allocation for Project Saffron was logged on 2026-06-25."}, {"sent_id": 9, "text": "Node Pine-07 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 096"}
{"doc_id": "doc_000097", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone Z1 for Project Anchor was revised on 2026-04-21."}, {"sent_id": 3, "text": "The schedule moved from 2026-04-29 to 2026-05-03."}, {"sent_id": 4, "text": "Project Anchor received 4 GPUs from Node Spruce-03 in the 2026-04-10 allocation log."}, {"sent_id": 5, "text": "Project Aster ran Helix-3B on OrionBench-2 using Node Poplar-12 on 2026-05-19."}, {"sent_id": 6, "text": "The reported macro F1 was 0.542."}, {"sent_id": 7, "text": "The staffing note for Project Saffron was dated 2026-06-09."}, {"sent_id": 8, "text": "It named Petra Gray for the evaluation owner duties."}, {"sent_id": 9, "text": "For Project Saffron, the run using Vela-8B on Node Fir-10 did not complete on 2026-05-01; the issue was a checkpoint-mismatch error."}, {"sent_id": 10, "text": "Milestone F1 for Project Saffron was revised on 2026-06-23."}, {"sent_id": 11, "text": "The schedule moved from 2026-07-11 to 2026-07-17."}, {"sent_id": 12, "text": "The 2026-04-21 result for Project Anchor put Orchid-7B on LumenFacts at 0.695 latency efficiency score, with Node Cedar-02 handling the run."}], "source": "synthetic_rewritten", "title": "Project Update 097"}
{"doc_id": "doc_000098", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-26 failure note says Lumen-8B on Node Spruce-03 failed for Project Aster because of an unstable-validation-loss error."}, {"sent_id": 3, "text": "For Project Sonata, the run using Mira-8B on Node Sycamore-13 did not complete on 2026-06-04; the issue was an out-of-memory error."}, {"sent_id": 4, "text": "Project Saffron had a failed run on 2026-05-01 involving Mira-7B and Node Sycamore-13."}, {"sent_id": 5, "text": "The recorded cause was a missing-index error."}, {"sent_id": 6, "text": "For reranking, the 2026-06-12 note says Project Saffron went with Vela-8B."}, {"sent_id": 7, "text": "Project Anchor ran River-7B on SignalSet-2 using Node Rowan-09 on 2026-04-20."}, {"sent_id": 8, "text": "The reported latency efficiency score was 0.869."}, {"sent_id": 9, "text": "Vela-7B was the model selected by Project Sonata on 2026-06-03 for the reranking track."}, {"sent_id": 10, "text": "Project Sonata ran Vela-7B on LabQA-2 using Node Fir-10 on 2026-06-30."}, {"sent_id": 11, "text": "The reported evidence F1 was 0.670."}], "source": "synthetic_rewritten", "title": "Lab Digest 098"}
{"doc_id": "doc_000099", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Sonata's Milestone D2 deadline change on 2026-04-24 replaced 2026-05-09 with 2026-05-05."}, {"sent_id": 3, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-06-21."}, {"sent_id": 4, "text": "The chosen model was Atlas-3B."}, {"sent_id": 5, "text": "For reranking, the 2026-05-20 note says Project Aster went with Orchid-7B."}, {"sent_id": 6, "text": "A compute allocation for Project Nereid was logged on 2026-05-09."}, {"sent_id": 7, "text": "Node Aspen-01 supplied 1 GPU."}, {"sent_id": 8, "text": "Project Anchor had a method update on 2026-06-21."}, {"sent_id": 9, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 10, "text": "Project Sonata ran Vela-7B on LabQA-2 using Node Fir-10 on 2026-05-10."}, {"sent_id": 11, "text": "The reported macro F1 was 0.730."}, {"sent_id": 12, "text": "Aster-3B was the model selected by Project Sonata on 2026-05-20 for the evidence retrieval track."}, {"sent_id": 13, "text": "The staffing note for Project Anchor was dated 2026-05-19."}, {"sent_id": 14, "text": "It named Dr. Arun Bauer for the evaluation owner duties."}], "source": "synthetic_rewritten", "title": "Project Update 099"}
{"doc_id": "doc_000100", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The staffing note for Project Aster was dated 2026-04-27."}, {"sent_id": 3, "text": "It named Mina Shah for the lead duties."}, {"sent_id": 4, "text": "On 2026-06-04, Project Sonata shifted Milestone X1: 2026-06-21 became 2026-06-30."}, {"sent_id": 5, "text": "For Project Saffron, the run using Mira-8B on Node Sycamore-13 did not complete on 2026-04-30; the issue was a checkpoint-mismatch error."}, {"sent_id": 6, "text": "In the 2026-06-29 staffing section, Project Saffron listed Lena Costa as the data steward."}, {"sent_id": 7, "text": "In the 2026-05-03 staffing section, Project Meridian listed Elian Ames as the evaluation owner."}, {"sent_id": 8, "text": "A compute allocation for Project Meridian was logged on 2026-05-23."}, {"sent_id": 9, "text": "Node Hazel-14 supplied 4 GPUs."}, {"sent_id": 10, "text": "Project Sonata had a method update on 2026-05-03."}, {"sent_id": 11, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 12, "text": "The 2026-06-20 method entry for Project Saffron moved from BM25 retrieval to evidence pooling."}, {"sent_id": 13, "text": "Atlas-7B was the model selected by Project Nereid on 2026-04-05 for the evidence retrieval track."}], "source": "synthetic_rewritten", "title": "Review Packet 100"}
{"doc_id": "doc_000101", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Sonata was logged on 2026-04-22."}, {"sent_id": 3, "text": "Node Sycamore-13 supplied 3 GPUs."}, {"sent_id": 4, "text": "Project Anchor's Milestone N1 deadline change on 2026-04-07 replaced 2026-04-29 with 2026-05-05."}, {"sent_id": 5, "text": "Lumen-3B was the model selected by Project Anchor on 2026-06-11 for the calibration track."}, {"sent_id": 6, "text": "Using Node Hazel-14, Atlas-8B produced 0.714 evidence F1 on RiverBench-2 for Project Nereid on 2026-04-25."}, {"sent_id": 7, "text": "The review note for Pale Compass Study 3 was dated 2026-05-04."}, {"sent_id": 8, "text": "It described the method as reward reranking and said it did not use a reward model."}, {"sent_id": 9, "text": "In the 2026-06-01 staffing section, Project Meridian listed Mira Sato as the data steward."}, {"sent_id": 10, "text": "In the 2026-05-04 staffing section, Project Meridian listed Nadia Singh as the data steward."}, {"sent_id": 11, "text": "A compute allocation for Project Anchor was logged on 2026-06-17."}, {"sent_id": 12, "text": "Node Juniper-06 supplied 4 GPUs."}, {"sent_id": 13, "text": "In the 2026-05-26 staffing section, Project Nereid listed Nora Sol as the retrieval owner."}], "source": "synthetic_rewritten", "title": "Run Brief 101"}
{"doc_id": "doc_000102", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-06-10 method entry for Project Nereid moved from BM25 retrieval to evidence pooling."}, {"sent_id": 3, "text": "Nested Verifier Study 3 appeared in a 2026-05-20 review entry using hard-negative mining; the note said it did not use a reward model."}, {"sent_id": 4, "text": "Project Meridian had a failed run on 2026-05-01 involving Atlas-7B and Node Hazel-14."}, {"sent_id": 5, "text": "The recorded cause was a missing-index error."}, {"sent_id": 6, "text": "Project Sonata had a failed run on 2026-04-24 involving Aster-7B and Node Willow-05."}, {"sent_id": 7, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 8, "text": "Using Node Spruce-03, Lumen-8B produced 0.893 latency efficiency score on MemoTrace-3 for Project Anchor on 2026-06-27."}, {"sent_id": 9, "text": "A compute allocation for Project Nereid was logged on 2026-06-13."}, {"sent_id": 10, "text": "Node Hazel-14 supplied 3 GPUs."}], "source": "synthetic_rewritten", "title": "Review Packet 102"}
{"doc_id": "doc_000103", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Milestone X1 for Project Sonata was revised on 2026-06-05."}, {"sent_id": 3, "text": "The schedule moved from 2026-06-12 to 2026-06-11."}, {"sent_id": 4, "text": "The 2026-04-20 result for Project Nereid put Cedar-8B on NereidNotes-3 at 0.726 evidence F1, with Node Laurel-11 handling the run."}, {"sent_id": 5, "text": "Aster-7B was the model selected by Project Sonata on 2026-06-11 for the evidence retrieval track."}, {"sent_id": 6, "text": "In the 2026-05-24 staffing section, Project Nereid listed Vera Torres as the retrieval owner."}, {"sent_id": 7, "text": "Milestone D1 for Project Meridian was revised on 2026-05-05."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-17 to 2026-05-21."}, {"sent_id": 9, "text": "Using Node Juniper-06, Finch-3B produced 0.521 accuracy on VestaLogs for Project Anchor on 2026-06-13."}], "source": "synthetic_rewritten", "title": "Run Brief 103"}
{"doc_id": "doc_000104", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-06-02 staffing section, Project Meridian listed Elian Ames as the evaluation owner."}, {"sent_id": 3, "text": "Project Nereid had a method update on 2026-04-29."}, {"sent_id": 4, "text": "The old approach was confidence calibration, and the replacement was chain verification."}, {"sent_id": 5, "text": "Project Aster received 1 GPU from Node Poplar-12 in the 2026-05-23 allocation log."}, {"sent_id": 6, "text": "On 2026-06-04, Project Nereid shifted Milestone B2: 2026-06-23 became 2026-06-20."}, {"sent_id": 7, "text": "On 2026-04-05, Project Saffron retired contrastive tuning in favor of threshold search."}, {"sent_id": 8, "text": "Using Node Fir-10, Vela-8B produced 0.610 macro F1 on LabQA-2 for Project Sonata on 2026-06-29."}], "source": "synthetic_rewritten", "title": "Review Packet 104"}
{"doc_id": "doc_000105", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "A compute allocation for Project Aster was logged on 2026-05-09."}, {"sent_id": 3, "text": "Node Rowan-09 supplied 3 GPUs."}, {"sent_id": 4, "text": "Project Aster ran Finch-3B on NereidNotes-2 using Node Juniper-06 on 2026-06-09."}, {"sent_id": 5, "text": "The reported evidence F1 was 0.866."}, {"sent_id": 6, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-14."}, {"sent_id": 7, "text": "The chosen model was Cedar-8B."}, {"sent_id": 8, "text": "Project Anchor's Milestone Z1 deadline change on 2026-04-14 replaced 2026-05-02 with 2026-05-08."}, {"sent_id": 9, "text": "Atlas-3B was the model selected by Project Meridian on 2026-04-29 for the evidence retrieval track."}, {"sent_id": 10, "text": "Kestrel-7B was the model selected by Project Nereid on 2026-05-31 for the claim classification track."}, {"sent_id": 11, "text": "Vela-8B was the model selected by Project Saffron on 2026-06-16 for the reranking track."}, {"sent_id": 12, "text": "The staffing note for Project Saffron was dated 2026-05-04."}, {"sent_id": 13, "text": "It named Petra Adler for the data steward duties."}, {"sent_id": 14, "text": "Using Node Poplar-12, Helix-3B produced 0.812 evidence F1 on OrionBench-2 for Project Aster on 2026-05-04."}], "source": "synthetic_rewritten", "title": "Run Brief 105"}
{"doc_id": "doc_000106", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for calibration on 2026-04-15."}, {"sent_id": 3, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 4, "text": "Project Saffron logged a personnel update on 2026-05-19: Dr. Theo Lind took the evaluation owner responsibilities."}, {"sent_id": 5, "text": "Project Saffron ran Aster-8B on SignalSet using Node Willow-05 on 2026-06-09."}, {"sent_id": 6, "text": "The reported latency efficiency score was 0.619."}, {"sent_id": 7, "text": "Project Meridian's Milestone J1 deadline change on 2026-06-29 replaced 2026-07-19 with 2026-07-27."}, {"sent_id": 8, "text": "Project Meridian had a method update on 2026-04-15."}, {"sent_id": 9, "text": "The old approach was alias expansion, and the replacement was dense retrieval."}, {"sent_id": 10, "text": "Project Nereid had a failed run on 2026-04-10 involving Cedar-3B and Node Laurel-11."}, {"sent_id": 11, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 12, "text": "Milestone B2 for Project Nereid was revised on 2026-05-28."}, {"sent_id": 13, "text": "The schedule moved from 2026-06-06 to 2026-06-15."}], "source": "synthetic_rewritten", "title": "Review Packet 106"}
{"doc_id": "doc_000107", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian had a method update on 2026-05-20."}, {"sent_id": 3, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 4, "text": "In the 2026-04-20 staffing section, Project Saffron listed Priya Vale as the evaluation owner."}, {"sent_id": 5, "text": "In the 2026-04-13 staffing section, Project Sonata listed Rohan Singh as the retrieval owner."}, {"sent_id": 6, "text": "Mira-3B was the model selected by Project Sonata on 2026-05-13 for the calibration track."}, {"sent_id": 7, "text": "Project Nereid ran Atlas-3B on RiverBench-2 using Node Hazel-14 on 2026-04-18."}, {"sent_id": 8, "text": "The reported macro F1 was 0.684."}, {"sent_id": 9, "text": "Project Meridian ran Nova-7B on LumenFacts-2 using Node Aspen-01 on 2026-05-26."}, {"sent_id": 10, "text": "The reported latency efficiency score was 0.585."}, {"sent_id": 11, "text": "On 2026-04-01, Project Anchor shifted Milestone B1: 2026-04-27 became 2026-05-03."}, {"sent_id": 12, "text": "Project Aster had a failed run on 2026-05-07 involving Helix-3B and Node Poplar-12."}, {"sent_id": 13, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 14, "text": "Project Aster's Milestone T1 deadline change on 2026-04-07 replaced 2026-04-22 with 2026-04-18."}], "source": "synthetic_rewritten", "title": "Run Brief 107"}
{"doc_id": "doc_000108", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-08, Project Saffron retired confidence calibration in favor of chain verification."}, {"sent_id": 3, "text": "Milestone L1 for Project Sonata was revised on 2026-05-08."}, {"sent_id": 4, "text": "The schedule moved from 2026-05-29 to 2026-06-07."}, {"sent_id": 5, "text": "A compute allocation for Project Anchor was logged on 2026-06-20."}, {"sent_id": 6, "text": "Node Cedar-02 supplied 4 GPUs."}, {"sent_id": 7, "text": "Using Node Laurel-11, Cedar-8B produced 0.666 evidence F1 on NereidNotes-3 for Project Nereid on 2026-05-02."}, {"sent_id": 8, "text": "Using Node Pine-07, Quartz-7B produced 0.622 macro F1 on OrionBench for Project Sonata on 2026-05-16."}, {"sent_id": 9, "text": "Project Anchor logged a personnel update on 2026-05-12: Leo Park took the evaluation owner responsibilities."}, {"sent_id": 10, "text": "Project Anchor had a method update on 2026-04-22."}, {"sent_id": 11, "text": "The old approach was contrastive tuning, and the replacement was threshold search."}, {"sent_id": 12, "text": "Project Sonata recorded a model choice for evidence retrieval on 2026-06-09."}, {"sent_id": 13, "text": "The chosen model was Aster-7B."}], "source": "synthetic_rewritten", "title": "Review Packet 108"}
{"doc_id": "doc_000109", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian's Milestone V1 deadline change on 2026-05-22 replaced 2026-06-07 with 2026-06-11."}, {"sent_id": 3, "text": "Project Nereid had a failed run on 2026-05-01 involving Cedar-3B and Node Laurel-11."}, {"sent_id": 4, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 5, "text": "Project Aster recorded a model choice for calibration on 2026-06-17."}, {"sent_id": 6, "text": "The chosen model was Lumen-8B."}, {"sent_id": 7, "text": "On 2026-04-20, the Stable Chains Study 1 review reported chain verification and said it did not use a reward model."}, {"sent_id": 8, "text": "Project Meridian had a method update on 2026-05-31."}, {"sent_id": 9, "text": "The old approach was LoRA adaptation, and the replacement was alias expansion."}, {"sent_id": 10, "text": "Project Meridian had a failed run on 2026-06-19 involving Nimbus-8B and Node Elm-08."}, {"sent_id": 11, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 12, "text": "Milestone F1 for Project Sonata was revised on 2026-06-26."}, {"sent_id": 13, "text": "The schedule moved from 2026-07-19 to 2026-07-17."}], "source": "synthetic_rewritten", "title": "Run Brief 109"}
{"doc_id": "doc_000110", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Birch-04, Kestrel-3B produced 0.747 latency efficiency score on CedarQA for Project Meridian on 2026-05-30."}, {"sent_id": 3, "text": "On 2026-06-24, Project Aster shifted Milestone H1: 2026-07-03 became 2026-07-12."}, {"sent_id": 4, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-04-30."}, {"sent_id": 5, "text": "The chosen model was Atlas-3B."}, {"sent_id": 6, "text": "On 2026-06-16, Project Nereid shifted Milestone D1: 2026-07-03 became 2026-07-08."}, {"sent_id": 7, "text": "Project Saffron recorded a model choice for error analysis on 2026-05-15."}, {"sent_id": 8, "text": "The chosen model was Marble-3B."}, {"sent_id": 9, "text": "Mira-8B was the model selected by Project Saffron on 2026-05-21 for the calibration track."}], "source": "synthetic_rewritten", "title": "Lab Digest 110"}
{"doc_id": "doc_000111", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Anchor ran Orchid-3B on LumenFacts using Node Cedar-02 on 2026-04-18."}, {"sent_id": 3, "text": "The reported latency efficiency score was 0.515."}, {"sent_id": 4, "text": "Project Anchor ran River-8B on SignalSet-2 using Node Rowan-09 on 2026-04-13."}, {"sent_id": 5, "text": "The reported accuracy was 0.839."}, {"sent_id": 6, "text": "Project Saffron logged a personnel update on 2026-06-02: Petra Adler took the data steward responsibilities."}, {"sent_id": 7, "text": "In the 2026-05-26 staffing section, Project Sonata listed Dr. Jonas Nolan as the retrieval owner."}, {"sent_id": 8, "text": "For Project Anchor, the run using Finch-8B on Node Juniper-06 did not complete on 2026-04-30; the issue was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "Project Nereid had a method update on 2026-04-25."}, {"sent_id": 10, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 11, "text": "Using Node Laurel-11, Cedar-7B produced 0.786 evidence F1 on NereidNotes-3 for Project Nereid on 2026-05-30."}], "source": "synthetic_rewritten", "title": "Project Update 111"}
{"doc_id": "doc_000112", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The staffing note for Project Sonata was dated 2026-05-04."}, {"sent_id": 3, "text": "It named Mara Lane for the retrieval owner duties."}, {"sent_id": 4, "text": "Project Anchor received 4 GPUs from Node Spruce-03 in the 2026-04-06 allocation log."}, {"sent_id": 5, "text": "The 2026-06-05 failure note says Mira-7B on Node Sycamore-13 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 6, "text": "Milestone L1 for Project Sonata was revised on 2026-05-14."}, {"sent_id": 7, "text": "The schedule moved from 2026-06-04 to 2026-06-09."}, {"sent_id": 8, "text": "In the 2026-05-04 staffing section, Project Anchor listed Kira Frost as the data steward."}, {"sent_id": 9, "text": "The 2026-06-03 method entry for Project Anchor moved from LoRA adaptation to alias expansion."}, {"sent_id": 10, "text": "Project Meridian logged a personnel update on 2026-06-29: Mira Nolan took the evaluation owner responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 112"}
{"doc_id": "doc_000113", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For Project Sonata, the run using Aster-7B on Node Willow-05 did not complete on 2026-04-30; the issue was an unstable-validation-loss error."}, {"sent_id": 3, "text": "Project Aster recorded a model choice for error analysis on 2026-06-28."}, {"sent_id": 4, "text": "The chosen model was Finch-8B."}, {"sent_id": 5, "text": "Using Node Elm-08, Nimbus-8B produced 0.708 evidence F1 on TraceEval-2 for Project Nereid on 2026-04-21."}, {"sent_id": 6, "text": "On 2026-06-17, Project Nereid retired alias expansion in favor of dense retrieval."}, {"sent_id": 7, "text": "Project Sonata received 3 GPUs from Node Maple-01 in the 2026-04-01 allocation log."}, {"sent_id": 8, "text": "Project Meridian received 4 GPUs from Node Hazel-14 in the 2026-06-29 allocation log."}], "source": "synthetic_rewritten", "title": "Project Update 113"}
{"doc_id": "doc_000114", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-06-14 staffing section, Project Nereid listed Evan Iyer as the retrieval owner."}, {"sent_id": 3, "text": "Project Saffron had a failed run on 2026-06-11 involving Mira-3B and Node Sycamore-13."}, {"sent_id": 4, "text": "The recorded cause was a missing-index error."}, {"sent_id": 5, "text": "Cedar-3B was the model selected by Project Meridian on 2026-04-12 for the error analysis track."}, {"sent_id": 6, "text": "Project Meridian had a failed run on 2026-06-05 involving Nimbus-7B and Node Elm-08."}, {"sent_id": 7, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 8, "text": "A compute allocation for Project Saffron was logged on 2026-06-05."}, {"sent_id": 9, "text": "Node Sycamore-13 supplied 4 GPUs."}, {"sent_id": 10, "text": "Project Saffron's Milestone L1 deadline change on 2026-05-06 replaced 2026-05-22 with 2026-05-30."}, {"sent_id": 11, "text": "On 2026-04-25, Project Saffron retired rank fusion in favor of teacher distillation."}, {"sent_id": 12, "text": "The 2026-06-06 compute note links Node Juniper-06 to Project Aster with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Lab Digest 114"}
{"doc_id": "doc_000115", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-06-03 compute note links Node Rowan-09 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 3, "text": "On 2026-04-06, Project Anchor shifted Milestone N1: 2026-04-18 became 2026-04-22."}, {"sent_id": 4, "text": "Project Saffron recorded a model choice for error analysis on 2026-06-10."}, {"sent_id": 5, "text": "The chosen model was Marble-8B."}, {"sent_id": 6, "text": "In the 2026-06-16 staffing section, Project Aster listed Talia Marin as the lead."}, {"sent_id": 7, "text": "Milestone D1 for Project Meridian was revised on 2026-06-18."}, {"sent_id": 8, "text": "The schedule moved from 2026-06-30 to 2026-07-04."}, {"sent_id": 9, "text": "Project Saffron had a failed run on 2026-05-07 involving Vela-8B and Node Fir-10."}, {"sent_id": 10, "text": "The recorded cause was a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Project Update 115"}
{"doc_id": "doc_000116", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor logged a personnel update on 2026-04-07: Leo Park took the evaluation owner responsibilities."}, {"sent_id": 3, "text": "Project Nereid had a method update on 2026-05-07."}, {"sent_id": 4, "text": "The old approach was contrastive tuning, and the replacement was threshold search."}, {"sent_id": 5, "text": "On 2026-04-11, Project Sonata retired confidence calibration in favor of chain verification."}, {"sent_id": 6, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-04-15."}, {"sent_id": 7, "text": "The chosen model was Atlas-8B."}, {"sent_id": 8, "text": "The 2026-04-19 method entry for Project Anchor moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 9, "text": "Using Node Spruce-03, Lumen-7B produced 0.833 latency efficiency score on MemoTrace-3 for Project Anchor on 2026-06-13."}, {"sent_id": 10, "text": "Project Saffron had a failed run on 2026-05-16 involving Marble-3B and Node Maple-01."}, {"sent_id": 11, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 12, "text": "The staffing note for Project Anchor was dated 2026-04-20."}, {"sent_id": 13, "text": "It named Iris Lane for the evaluation owner duties."}, {"sent_id": 14, "text": "Project Sonata received 3 GPUs from Node Pine-07 in the 2026-06-13 allocation log."}], "source": "synthetic_rewritten", "title": "Lab Digest 116"}
{"doc_id": "doc_000117", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For claim classification, the 2026-06-14 note says Project Anchor went with Helix-3B."}, {"sent_id": 3, "text": "Cedar-8B was the model selected by Project Nereid on 2026-04-05 for the error analysis track."}, {"sent_id": 4, "text": "Project Anchor received 4 GPUs from Node Spruce-03 in the 2026-05-16 allocation log."}, {"sent_id": 5, "text": "Project Aster had a method update on 2026-05-03."}, {"sent_id": 6, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 7, "text": "For Project Sonata, the run using Marble-7B on Node Maple-01 did not complete on 2026-04-04; the issue was an out-of-memory error."}, {"sent_id": 8, "text": "Project Aster received 1 GPU from Node Rowan-09 in the 2026-06-18 allocation log."}], "source": "synthetic_rewritten", "title": "Project Update 117"}
{"doc_id": "doc_000118", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone P1 for Project Nereid was revised on 2026-04-15."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-08 to 2026-05-06."}, {"sent_id": 4, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-15."}, {"sent_id": 5, "text": "The chosen model was Cedar-7B."}, {"sent_id": 6, "text": "Project Anchor had a failed run on 2026-04-09 involving Helix-8B and Node Poplar-12."}, {"sent_id": 7, "text": "The recorded cause was a missing-index error."}, {"sent_id": 8, "text": "Milestone X1 for Project Sonata was revised on 2026-05-06."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-23 to 2026-05-28."}, {"sent_id": 10, "text": "Project Nereid recorded a model choice for reranking on 2026-06-23."}, {"sent_id": 11, "text": "The chosen model was Nova-8B."}, {"sent_id": 12, "text": "Project Anchor's Milestone N1 deadline change on 2026-05-22 replaced 2026-06-03 with 2026-06-11."}], "source": "synthetic_rewritten", "title": "Lab Digest 118"}
{"doc_id": "doc_000119", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-22 result for Project Saffron put Aster-7B on SignalSet at 0.619 accuracy, with Node Willow-05 handling the run."}, {"sent_id": 3, "text": "The 2026-06-14 result for Project Aster put Orchid-3B on LabQA-3 at 0.620 macro F1, with Node Cedar-02 handling the run."}, {"sent_id": 4, "text": "On 2026-05-08, Project Saffron shifted Milestone F1: 2026-05-16 became 2026-05-24."}, {"sent_id": 5, "text": "Project Saffron's Milestone D2 deadline change on 2026-04-29 replaced 2026-05-19 with 2026-05-27."}, {"sent_id": 6, "text": "Using Node Juniper-06, Finch-7B produced 0.896 evidence F1 on NereidNotes-2 for Project Aster on 2026-06-29."}, {"sent_id": 7, "text": "Project Meridian recorded a model choice for claim classification on 2026-05-29."}, {"sent_id": 8, "text": "The chosen model was Kestrel-7B."}, {"sent_id": 9, "text": "For claim classification, the 2026-05-14 note says Project Meridian went with Kestrel-7B."}, {"sent_id": 10, "text": "For evidence retrieval, the 2026-05-31 note says Project Aster went with River-7B."}], "source": "synthetic_rewritten", "title": "Project Update 119"}
{"doc_id": "doc_000120", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-04-24."}, {"sent_id": 3, "text": "The chosen model was Aster-8B."}, {"sent_id": 4, "text": "For calibration, the 2026-04-30 note says Project Aster went with Lumen-7B."}, {"sent_id": 5, "text": "The staffing note for Project Saffron was dated 2026-05-25."}, {"sent_id": 6, "text": "It named Lena Costa for the data steward duties."}, {"sent_id": 7, "text": "The 2026-04-08 method entry for Project Nereid moved from reward reranking to data mixing."}, {"sent_id": 8, "text": "The 2026-05-02 method entry for Project Anchor moved from threshold search to reward reranking."}, {"sent_id": 9, "text": "Project Saffron recorded a model choice for calibration on 2026-05-31."}, {"sent_id": 10, "text": "The chosen model was Mira-8B."}, {"sent_id": 11, "text": "Project Meridian received 2 GPUs from Node Elm-08 in the 2026-05-08 allocation log."}], "source": "synthetic_rewritten", "title": "Run Brief 120"}
{"doc_id": "doc_000121", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-05-01, the Quiet Retriever Study 4 review reported sentence pruning and said it did not use a reward model."}, {"sent_id": 3, "text": "Project Saffron's Milestone D2 deadline change on 2026-04-27 replaced 2026-05-07 with 2026-05-13."}, {"sent_id": 4, "text": "Project Meridian had a failed run on 2026-06-25 involving Nova-7B and Node Aspen-01."}, {"sent_id": 5, "text": "The recorded cause was a missing-index error."}, {"sent_id": 6, "text": "Project Anchor received 2 GPUs from Node Juniper-06 in the 2026-05-21 allocation log."}, {"sent_id": 7, "text": "Mira-7B was the model selected by Project Sonata on 2026-06-14 for the calibration track."}, {"sent_id": 8, "text": "Project Aster recorded a model choice for calibration on 2026-06-10."}, {"sent_id": 9, "text": "The chosen model was Lumen-3B."}, {"sent_id": 10, "text": "The 2026-05-22 failure note says Lumen-8B on Node Spruce-03 failed for Project Anchor because of a missing-index error."}, {"sent_id": 11, "text": "The 2026-06-06 result for Project Meridian put Atlas-7B on SignalSet-3 at 0.699 latency efficiency score, with Node Hazel-14 handling the run."}, {"sent_id": 12, "text": "Marble-8B was the model selected by Project Sonata on 2026-06-11 for the error analysis track."}], "source": "synthetic_rewritten", "title": "Review Packet 121"}
{"doc_id": "doc_000122", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-05-25 staffing section, Project Anchor listed Iris Lane as the evaluation owner."}, {"sent_id": 3, "text": "Project Sonata logged a personnel update on 2026-04-06: Rohan Park took the lead responsibilities."}, {"sent_id": 4, "text": "For error analysis, the 2026-04-10 note says Project Meridian went with Cedar-7B."}, {"sent_id": 5, "text": "A compute allocation for Project Nereid was logged on 2026-04-10."}, {"sent_id": 6, "text": "Node Aspen-01 supplied 3 GPUs."}, {"sent_id": 7, "text": "Project Sonata logged a personnel update on 2026-06-22: Samir Ames took the lead responsibilities."}, {"sent_id": 8, "text": "On 2026-04-26, Project Aster retired QLoRA adaptation in favor of hybrid retrieval."}], "source": "synthetic_rewritten", "title": "Run Brief 122"}
{"doc_id": "doc_000123", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Vela-8B was the model selected by Project Saffron on 2026-05-28 for the reranking track."}, {"sent_id": 3, "text": "Project Aster had a failed run on 2026-04-17 involving Orchid-3B and Node Cedar-02."}, {"sent_id": 4, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 5, "text": "Project Aster received 1 GPU from Node Cedar-02 in the 2026-06-20 allocation log."}, {"sent_id": 6, "text": "For Project Nereid, the run using Nova-7B on Node Aspen-01 did not complete on 2026-05-15; the issue was an out-of-memory error."}, {"sent_id": 7, "text": "Project Saffron received 2 GPUs from Node Willow-05 in the 2026-04-29 allocation log."}, {"sent_id": 8, "text": "Project Aster received 3 GPUs from Node Juniper-06 in the 2026-06-27 allocation log."}, {"sent_id": 9, "text": "For Project Nereid, the run using Atlas-8B on Node Hazel-14 did not complete on 2026-06-19; the issue was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Review Packet 123"}
{"doc_id": "doc_000124", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The review note for Pale Compass Study 1 was dated 2026-06-08."}, {"sent_id": 3, "text": "It described the method as contrastive tuning and said it did not use a reward model."}, {"sent_id": 4, "text": "Using Node Sycamore-13, Mira-7B produced 0.583 latency efficiency score on MemoTrace-2 for Project Saffron on 2026-04-07."}, {"sent_id": 5, "text": "The 2026-04-12 method entry for Project Meridian moved from rank fusion to teacher distillation."}, {"sent_id": 6, "text": "The staffing note for Project Aster was dated 2026-05-31."}, {"sent_id": 7, "text": "It named Talia Reed for the retrieval owner duties."}, {"sent_id": 8, "text": "Using Node Laurel-11, Cedar-7B produced 0.711 latency efficiency score on VestaLogs-2 for Project Meridian on 2026-06-01."}, {"sent_id": 9, "text": "On 2026-05-14, Project Meridian shifted Milestone J1: 2026-05-24 became 2026-06-03."}], "source": "synthetic_rewritten", "title": "Run Brief 124"}
{"doc_id": "doc_000125", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Finch-3B was the model selected by Project Aster on 2026-04-03 for the error analysis track."}, {"sent_id": 3, "text": "In the 2026-05-19 staffing section, Project Meridian listed Camila Brooks as the data steward."}, {"sent_id": 4, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-04-24."}, {"sent_id": 5, "text": "The chosen model was Atlas-3B."}, {"sent_id": 6, "text": "The staffing note for Project Anchor was dated 2026-06-01."}, {"sent_id": 7, "text": "It named Selene Kim for the evaluation owner duties."}, {"sent_id": 8, "text": "The staffing note for Project Sonata was dated 2026-05-10."}, {"sent_id": 9, "text": "It named Samir Kwan for the retrieval owner duties."}, {"sent_id": 10, "text": "The 2026-06-26 compute note links Node Sycamore-13 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 11, "text": "The staffing note for Project Nereid was dated 2026-05-04."}, {"sent_id": 12, "text": "It named Nora Sol for the retrieval owner duties."}], "source": "synthetic_rewritten", "title": "Review Packet 125"}
{"doc_id": "doc_000126", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Nimbus-3B was the model selected by Project Nereid on 2026-06-04 for the calibration track."}, {"sent_id": 3, "text": "In the 2026-06-30 staffing section, Project Meridian listed Mira Sato as the data steward."}, {"sent_id": 4, "text": "On 2026-05-17, Project Sonata retired BM25 retrieval in favor of evidence pooling."}, {"sent_id": 5, "text": "Project Nereid ran Atlas-8B on RiverBench-2 using Node Hazel-14 on 2026-04-07."}, {"sent_id": 6, "text": "The reported evidence F1 was 0.804."}, {"sent_id": 7, "text": "Cedar-3B was the model selected by Project Nereid on 2026-06-10 for the error analysis track."}, {"sent_id": 8, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-05-29."}, {"sent_id": 9, "text": "The chosen model was Atlas-3B."}, {"sent_id": 10, "text": "Using Node Sycamore-13, Mira-8B produced 0.643 accuracy on MemoTrace-2 for Project Saffron on 2026-05-30."}, {"sent_id": 11, "text": "Project Anchor logged a personnel update on 2026-05-04: Iris Lane took the evaluation owner responsibilities."}], "source": "synthetic_rewritten", "title": "Run Brief 126"}
{"doc_id": "doc_000127", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian's Milestone D1 deadline change on 2026-04-27 replaced 2026-05-09 with 2026-05-13."}, {"sent_id": 3, "text": "Project Sonata ran Vela-3B on LabQA-2 using Node Fir-10 on 2026-04-25."}, {"sent_id": 4, "text": "The reported macro F1 was 0.610."}, {"sent_id": 5, "text": "Project Saffron ran Aster-7B on SignalSet using Node Willow-05 on 2026-04-25."}, {"sent_id": 6, "text": "The reported latency efficiency score was 0.649."}, {"sent_id": 7, "text": "For Project Aster, the run using River-3B on Node Rowan-09 did not complete on 2026-05-29; the issue was an unstable-validation-loss error."}, {"sent_id": 8, "text": "For Project Meridian, the run using Nimbus-3B on Node Elm-08 did not complete on 2026-06-18; the issue was a missing-index error."}, {"sent_id": 9, "text": "Project Aster had a method update on 2026-05-17."}, {"sent_id": 10, "text": "The old approach was QLoRA adaptation, and the replacement was hybrid retrieval."}, {"sent_id": 11, "text": "The 2026-04-30 failure note says Cedar-7B on Node Laurel-11 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 12, "text": "Nested Verifier Study 2 appeared in a 2026-04-22 review entry using QLoRA adaptation; the note said it did not use a reward model."}, {"sent_id": 13, "text": "Project Sonata received 3 GPUs from Node Sycamore-13 in the 2026-04-20 allocation log."}], "source": "synthetic_rewritten", "title": "Review Packet 127"}
{"doc_id": "doc_000128", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-05-17, Project Nereid retired calibrated voting in favor of confidence calibration."}, {"sent_id": 3, "text": "The staffing note for Project Saffron was dated 2026-04-13."}, {"sent_id": 4, "text": "It named Owen Torres for the evaluation owner duties."}, {"sent_id": 5, "text": "The 2026-06-06 compute note links Node Spruce-03 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 6, "text": "The 2026-05-28 failure note says Cedar-8B on Node Laurel-11 failed for Project Meridian because of a missing-index error."}, {"sent_id": 7, "text": "A compute allocation for Project Meridian was logged on 2026-06-15."}, {"sent_id": 8, "text": "Node Birch-04 supplied 2 GPUs."}, {"sent_id": 9, "text": "For claim classification, the 2026-04-17 note says Project Nereid went with Kestrel-8B."}], "source": "synthetic_rewritten", "title": "Run Brief 128"}
{"doc_id": "doc_000129", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-04-25 result for Project Saffron put Marble-8B on VestaLogs-3 at 0.571 accuracy, with Node Maple-01 handling the run."}, {"sent_id": 3, "text": "The 2026-04-02 method entry for Project Aster moved from reward reranking to data mixing."}, {"sent_id": 4, "text": "For evidence retrieval, the 2026-06-04 note says Project Sonata went with Aster-8B."}, {"sent_id": 5, "text": "On 2026-06-17, Project Meridian shifted Milestone B2: 2026-07-11 became 2026-07-15."}, {"sent_id": 6, "text": "Project Meridian received 4 GPUs from Node Birch-04 in the 2026-06-20 allocation log."}, {"sent_id": 7, "text": "Project Anchor ran Helix-8B on CedarQA-3 using Node Poplar-12 on 2026-04-18."}, {"sent_id": 8, "text": "The reported accuracy was 0.827."}, {"sent_id": 9, "text": "A compute allocation for Project Meridian was logged on 2026-05-22."}, {"sent_id": 10, "text": "Node Birch-04 supplied 2 GPUs."}, {"sent_id": 11, "text": "On 2026-05-03, Project Meridian retired dense retrieval in favor of rank fusion."}], "source": "synthetic_rewritten", "title": "Review Packet 129"}
{"doc_id": "doc_000130", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata had a method update on 2026-06-24."}, {"sent_id": 3, "text": "The old approach was metric smoothing, and the replacement was hard-negative mining."}, {"sent_id": 4, "text": "On 2026-04-21, Project Nereid shifted Milestone B2: 2026-05-10 became 2026-05-07."}, {"sent_id": 5, "text": "A compute allocation for Project Aster was logged on 2026-04-13."}, {"sent_id": 6, "text": "Node Rowan-09 supplied 3 GPUs."}, {"sent_id": 7, "text": "Project Aster's Milestone N1 deadline change on 2026-05-19 replaced 2026-05-26 with 2026-05-25."}, {"sent_id": 8, "text": "The 2026-04-27 compute note links Node Willow-05 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 9, "text": "The staffing note for Project Nereid was dated 2026-04-21."}, {"sent_id": 10, "text": "It named Evan Moss for the lead duties."}, {"sent_id": 11, "text": "Project Meridian's Milestone V1 deadline change on 2026-05-29 replaced 2026-06-24 with 2026-07-04."}], "source": "synthetic_rewritten", "title": "Project Update 130"}
{"doc_id": "doc_000131", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-17 method entry for Project Saffron moved from LoRA adaptation to alias expansion."}, {"sent_id": 3, "text": "On 2026-06-21, Project Anchor retired threshold search in favor of reward reranking."}, {"sent_id": 4, "text": "For calibration, the 2026-04-09 note says Project Sonata went with Mira-3B."}, {"sent_id": 5, "text": "The staffing note for Project Nereid was dated 2026-05-05."}, {"sent_id": 6, "text": "It named Darian Hale for the retrieval owner duties."}, {"sent_id": 7, "text": "In the 2026-04-14 staffing section, Project Aster listed Talia Reed as the retrieval owner."}, {"sent_id": 8, "text": "The 2026-05-25 result for Project Aster put Helix-3B on OrionBench-2 at 0.512 macro F1, with Node Poplar-12 handling the run."}, {"sent_id": 9, "text": "Milestone T1 for Project Aster was revised on 2026-05-22."}, {"sent_id": 10, "text": "The schedule moved from 2026-06-16 to 2026-06-25."}], "source": "synthetic_rewritten", "title": "Lab Digest 131"}
{"doc_id": "doc_000132", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-27, Project Aster shifted Milestone B1: 2026-05-18 became 2026-05-27."}, {"sent_id": 3, "text": "Project Meridian's Milestone J1 deadline change on 2026-06-17 replaced 2026-06-27 with 2026-07-03."}, {"sent_id": 4, "text": "For claim classification, the 2026-05-08 note says Project Sonata went with Quartz-3B."}, {"sent_id": 5, "text": "The staffing note for Project Meridian was dated 2026-06-15."}, {"sent_id": 6, "text": "It named Dr. Ravi Reed for the evaluation owner duties."}, {"sent_id": 7, "text": "Milestone T1 for Project Anchor was revised on 2026-06-10."}, {"sent_id": 8, "text": "The schedule moved from 2026-06-20 to 2026-06-30."}, {"sent_id": 9, "text": "Project Meridian ran Nimbus-7B on MemoTrace using Node Elm-08 on 2026-06-29."}, {"sent_id": 10, "text": "The reported accuracy was 0.753."}, {"sent_id": 11, "text": "In the 2026-05-25 staffing section, Project Sonata listed Mara Lane as the retrieval owner."}, {"sent_id": 12, "text": "Project Sonata recorded a model choice for claim classification on 2026-04-30."}, {"sent_id": 13, "text": "The chosen model was Quartz-8B."}], "source": "synthetic_rewritten", "title": "Project Update 132"}
{"doc_id": "doc_000133", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata recorded a model choice for claim classification on 2026-05-19."}, {"sent_id": 3, "text": "The chosen model was Quartz-8B."}, {"sent_id": 4, "text": "For error analysis, the 2026-05-17 note says Project Sonata went with Marble-3B."}, {"sent_id": 5, "text": "Project Aster ran Helix-7B on OrionBench-2 using Node Poplar-12 on 2026-04-27."}, {"sent_id": 6, "text": "The reported macro F1 was 0.782."}, {"sent_id": 7, "text": "Project Nereid received 1 GPU from Node Hazel-14 in the 2026-05-13 allocation log."}, {"sent_id": 8, "text": "Project Saffron ran Quartz-8B on CedarQA-2 using Node Pine-07 on 2026-05-26."}, {"sent_id": 9, "text": "The reported accuracy was 0.637."}, {"sent_id": 10, "text": "Project Nereid logged a personnel update on 2026-06-02: Vera Torres took the retrieval owner responsibilities."}, {"sent_id": 11, "text": "On 2026-04-26, Project Saffron retired calibrated voting in favor of confidence calibration."}, {"sent_id": 12, "text": "Project Nereid recorded a model choice for error analysis on 2026-04-01."}, {"sent_id": 13, "text": "The chosen model was Cedar-3B."}, {"sent_id": 14, "text": "Project Sonata logged a personnel update on 2026-04-21: Sofia Nadir took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 133"}
{"doc_id": "doc_000134", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For Project Sonata, the run using Aster-8B on Node Willow-05 did not complete on 2026-05-08; the issue was an unstable-validation-loss error."}, {"sent_id": 3, "text": "A compute allocation for Project Aster was logged on 2026-04-24."}, {"sent_id": 4, "text": "Node Spruce-03 supplied 1 GPU."}, {"sent_id": 5, "text": "For calibration, the 2026-04-19 note says Project Aster went with Lumen-3B."}, {"sent_id": 6, "text": "Project Saffron's Milestone X1 deadline change on 2026-04-03 replaced 2026-04-15 with 2026-04-23."}, {"sent_id": 7, "text": "Project Meridian received 4 GPUs from Node Elm-08 in the 2026-04-09 allocation log."}, {"sent_id": 8, "text": "The 2026-04-18 compute note links Node Aspen-01 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 9, "text": "Milestone P1 for Project Meridian was revised on 2026-05-18."}, {"sent_id": 10, "text": "The schedule moved from 2026-05-26 to 2026-06-03."}], "source": "synthetic_rewritten", "title": "Project Update 134"}
{"doc_id": "doc_000135", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Pale Compass Study 3 appeared in a 2026-04-01 review entry using temporal filtering; the note said it did not use a reward model."}, {"sent_id": 3, "text": "The 2026-06-12 failure note says Orchid-3B on Node Cedar-02 failed for Project Anchor because of a missing-index error."}, {"sent_id": 4, "text": "Helix-3B was the model selected by Project Aster on 2026-06-18 for the claim classification track."}, {"sent_id": 5, "text": "On 2026-04-29, Project Aster shifted Milestone T1: 2026-05-24 became 2026-06-02."}, {"sent_id": 6, "text": "For Project Saffron, the run using Marble-7B on Node Maple-01 did not complete on 2026-05-21; the issue was a missing-index error."}, {"sent_id": 7, "text": "For Project Anchor, the run using Helix-8B on Node Poplar-12 did not complete on 2026-06-18; the issue was a checkpoint-mismatch error."}, {"sent_id": 8, "text": "A compute allocation for Project Nereid was logged on 2026-05-11."}, {"sent_id": 9, "text": "Node Hazel-14 supplied 1 GPU."}, {"sent_id": 10, "text": "The 2026-04-15 method entry for Project Meridian moved from metric smoothing to hard-negative mining."}], "source": "synthetic_rewritten", "title": "Lab Digest 135"}
{"doc_id": "doc_000136", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor logged a personnel update on 2026-05-12: Dr. Arun Kwan took the data steward responsibilities."}, {"sent_id": 3, "text": "Project Saffron received 4 GPUs from Node Maple-01 in the 2026-04-23 allocation log."}, {"sent_id": 4, "text": "On 2026-04-19, Project Nereid retired chain verification in favor of structured prompting."}, {"sent_id": 5, "text": "The 2026-06-28 method entry for Project Saffron moved from confidence calibration to chain verification."}, {"sent_id": 6, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-26."}, {"sent_id": 7, "text": "The chosen model was Cedar-3B."}, {"sent_id": 8, "text": "The 2026-04-18 failure note says Mira-7B on Node Sycamore-13 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 9, "text": "The 2026-04-12 result for Project Aster put River-8B on RiverBench at 0.584 macro F1, with Node Rowan-09 handling the run."}, {"sent_id": 10, "text": "Quiet Retriever Study 3 appeared in a 2026-05-27 review entry using metric smoothing; the note said it did not use a reward model."}, {"sent_id": 11, "text": "The staffing note for Project Nereid was dated 2026-04-06."}, {"sent_id": 12, "text": "It named Evan Moss for the lead duties."}], "source": "synthetic_rewritten", "title": "Project Update 136"}
{"doc_id": "doc_000137", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-03 compute note links Node Sycamore-13 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 3, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-04-21."}, {"sent_id": 4, "text": "The chosen model was Aster-3B."}, {"sent_id": 5, "text": "For Project Sonata, the run using Mira-3B on Node Sycamore-13 did not complete on 2026-05-28; the issue was an unstable-validation-loss error."}, {"sent_id": 6, "text": "Milestone B2 for Project Meridian was revised on 2026-04-16."}, {"sent_id": 7, "text": "The schedule moved from 2026-05-10 to 2026-05-18."}, {"sent_id": 8, "text": "On 2026-04-11, Project Aster retired calibrated voting in favor of confidence calibration."}, {"sent_id": 9, "text": "Using Node Poplar-12, Helix-8B produced 0.887 accuracy on CedarQA-3 for Project Anchor on 2026-04-06."}, {"sent_id": 10, "text": "For Project Aster, the run using Helix-7B on Node Poplar-12 did not complete on 2026-06-27; the issue was an out-of-memory error."}, {"sent_id": 11, "text": "Project Meridian received 2 GPUs from Node Elm-08 in the 2026-06-27 allocation log."}, {"sent_id": 12, "text": "On 2026-05-06, Project Saffron shifted Milestone D2: 2026-05-16 became 2026-05-22."}], "source": "synthetic_rewritten", "title": "Lab Digest 137"}
{"doc_id": "doc_000138", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-13 method entry for Project Sonata moved from cross-encoder reranking to late interaction."}, {"sent_id": 3, "text": "For Project Anchor, the run using Finch-7B on Node Juniper-06 did not complete on 2026-06-13; the issue was a missing-index error."}, {"sent_id": 4, "text": "On 2026-05-10, Project Saffron retired threshold search in favor of reward reranking."}, {"sent_id": 5, "text": "Vela-3B was the model selected by Project Saffron on 2026-05-03 for the reranking track."}, {"sent_id": 6, "text": "For Project Saffron, the run using Marble-8B on Node Maple-01 did not complete on 2026-04-10; the issue was a missing-index error."}, {"sent_id": 7, "text": "Project Sonata ran Quartz-3B on OrionBench using Node Pine-07 on 2026-05-24."}, {"sent_id": 8, "text": "The reported evidence F1 was 0.712."}, {"sent_id": 9, "text": "In the 2026-04-13 staffing section, Project Anchor listed Leo Park as the evaluation owner."}, {"sent_id": 10, "text": "For Project Anchor, the run using Finch-3B on Node Juniper-06 did not complete on 2026-06-20; the issue was a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Project Update 138"}
{"doc_id": "doc_000139", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Aster recorded a model choice for reranking on 2026-05-22."}, {"sent_id": 3, "text": "The chosen model was Orchid-7B."}, {"sent_id": 4, "text": "The 2026-04-03 failure note says Quartz-8B on Node Pine-07 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 5, "text": "For Project Anchor, the run using River-8B on Node Rowan-09 did not complete on 2026-04-16; the issue was a missing-index error."}, {"sent_id": 6, "text": "Project Anchor ran Finch-7B on VestaLogs using Node Juniper-06 on 2026-05-04."}, {"sent_id": 7, "text": "The reported accuracy was 0.851."}, {"sent_id": 8, "text": "Project Meridian ran Atlas-8B on SignalSet-3 using Node Hazel-14 on 2026-06-30."}, {"sent_id": 9, "text": "The reported accuracy was 0.579."}, {"sent_id": 10, "text": "The 2026-06-07 method entry for Project Nereid moved from temporal filtering to alias expansion."}, {"sent_id": 11, "text": "Nimbus-3B was the model selected by Project Meridian on 2026-04-28 for the calibration track."}, {"sent_id": 12, "text": "For claim classification, the 2026-05-12 note says Project Sonata went with Quartz-3B."}, {"sent_id": 13, "text": "On 2026-06-01, Project Aster shifted Milestone Z1: 2026-06-24 became 2026-06-22."}], "source": "synthetic_rewritten", "title": "Lab Digest 139"}
{"doc_id": "doc_000140", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata received 3 GPUs from Node Fir-10 in the 2026-05-15 allocation log."}, {"sent_id": 3, "text": "On 2026-05-06, Project Aster shifted Milestone B1: 2026-05-27 became 2026-06-05."}, {"sent_id": 4, "text": "The 2026-06-16 result for Project Nereid put Atlas-8B on RiverBench-2 at 0.714 macro F1, with Node Hazel-14 handling the run."}, {"sent_id": 5, "text": "Using Node Rowan-09, River-3B produced 0.884 evidence F1 on RiverBench for Project Aster on 2026-04-04."}, {"sent_id": 6, "text": "The 2026-06-28 method entry for Project Saffron moved from chain verification to structured prompting."}, {"sent_id": 7, "text": "Project Nereid ran Cedar-8B on NereidNotes-3 using Node Laurel-11 on 2026-06-02."}, {"sent_id": 8, "text": "The reported evidence F1 was 0.576."}, {"sent_id": 9, "text": "On 2026-06-07, Project Anchor retired calibrated voting in favor of confidence calibration."}, {"sent_id": 10, "text": "Orchid-3B was the model selected by Project Aster on 2026-05-14 for the reranking track."}, {"sent_id": 11, "text": "Using Node Elm-08, Nimbus-7B produced 0.558 macro F1 on TraceEval-2 for Project Nereid on 2026-06-16."}], "source": "synthetic_rewritten", "title": "Review Packet 140"}
{"doc_id": "doc_000141", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Meridian was logged on 2026-04-25."}, {"sent_id": 3, "text": "Node Hazel-14 supplied 4 GPUs."}, {"sent_id": 4, "text": "The 2026-05-02 compute note links Node Laurel-11 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 5, "text": "Project Meridian ran Nimbus-8B on MemoTrace using Node Elm-08 on 2026-04-25."}, {"sent_id": 6, "text": "The reported accuracy was 0.753."}, {"sent_id": 7, "text": "Project Saffron had a method update on 2026-06-21."}, {"sent_id": 8, "text": "The old approach was alias expansion, and the replacement was dense retrieval."}, {"sent_id": 9, "text": "For calibration, the 2026-05-24 note says Project Sonata went with Mira-7B."}, {"sent_id": 10, "text": "A compute allocation for Project Sonata was logged on 2026-05-16."}, {"sent_id": 11, "text": "Node Fir-10 supplied 1 GPU."}, {"sent_id": 12, "text": "The 2026-06-27 method entry for Project Anchor moved from hybrid retrieval to document chunking."}], "source": "synthetic_rewritten", "title": "Run Brief 141"}
{"doc_id": "doc_000142", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Aster logged a personnel update on 2026-04-07: Dr. Noah Vale took the lead responsibilities."}, {"sent_id": 3, "text": "The 2026-05-07 compute note links Node Aspen-01 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 4, "text": "Using Node Cedar-02, Orchid-3B produced 0.515 accuracy on LumenFacts for Project Anchor on 2026-06-09."}, {"sent_id": 5, "text": "For calibration, the 2026-05-10 note says Project Aster went with Lumen-7B."}, {"sent_id": 6, "text": "Project Meridian had a method update on 2026-05-20."}, {"sent_id": 7, "text": "The old approach was calibrated voting, and the replacement was confidence calibration."}, {"sent_id": 8, "text": "The staffing note for Project Anchor was dated 2026-06-08."}, {"sent_id": 9, "text": "It named Kira Iyer for the evaluation owner duties."}, {"sent_id": 10, "text": "On 2026-06-12, Project Saffron shifted Milestone L1: 2026-06-28 became 2026-07-02."}, {"sent_id": 11, "text": "Project Nereid logged a personnel update on 2026-05-04: Evan Iyer took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Review Packet 142"}
{"doc_id": "doc_000143", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Aster logged a personnel update on 2026-06-15: Mina Shah took the lead responsibilities."}, {"sent_id": 3, "text": "Nimbus-7B was the model selected by Project Nereid on 2026-04-01 for the calibration track."}, {"sent_id": 4, "text": "The 2026-05-27 compute note links Node Spruce-03 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 5, "text": "The staffing note for Project Sonata was dated 2026-05-11."}, {"sent_id": 6, "text": "It named Rohan Park for the lead duties."}, {"sent_id": 7, "text": "Project Aster ran River-8B on RiverBench using Node Rowan-09 on 2026-06-02."}, {"sent_id": 8, "text": "The reported evidence F1 was 0.524."}, {"sent_id": 9, "text": "The 2026-06-16 result for Project Anchor put Orchid-8B on LumenFacts at 0.545 latency efficiency score, with Node Cedar-02 handling the run."}], "source": "synthetic_rewritten", "title": "Run Brief 143"}
{"doc_id": "doc_000144", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For error analysis, the 2026-04-08 note says Project Anchor went with Finch-3B."}, {"sent_id": 3, "text": "On 2026-05-23, Project Saffron retired LoRA adaptation in favor of alias expansion."}, {"sent_id": 4, "text": "Project Meridian recorded a model choice for calibration on 2026-04-08."}, {"sent_id": 5, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 6, "text": "The 2026-05-15 compute note links Node Maple-01 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 7, "text": "Lumen-8B was the model selected by Project Anchor on 2026-06-12 for the calibration track."}, {"sent_id": 8, "text": "On 2026-06-17, Project Nereid retired temporal filtering in favor of alias expansion."}, {"sent_id": 9, "text": "The staffing note for Project Meridian was dated 2026-05-25."}, {"sent_id": 10, "text": "It named Mira Nolan for the evaluation owner duties."}, {"sent_id": 11, "text": "The staffing note for Project Meridian was dated 2026-06-02."}, {"sent_id": 12, "text": "It named Camila Brooks for the data steward duties."}], "source": "synthetic_rewritten", "title": "Review Packet 144"}
{"doc_id": "doc_000145", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For reranking, the 2026-05-14 note says Project Meridian went with Nova-3B."}, {"sent_id": 3, "text": "Using Node Rowan-09, River-8B produced 0.794 macro F1 on RiverBench for Project Aster on 2026-05-18."}, {"sent_id": 4, "text": "The staffing note for Project Aster was dated 2026-05-11."}, {"sent_id": 5, "text": "It named Anika Costa for the retrieval owner duties."}, {"sent_id": 6, "text": "On 2026-05-21, Project Nereid retired rank fusion in favor of teacher distillation."}, {"sent_id": 7, "text": "On 2026-05-14, Project Nereid shifted Milestone P1: 2026-06-06 became 2026-06-04."}, {"sent_id": 8, "text": "Project Saffron recorded a model choice for claim classification on 2026-04-22."}, {"sent_id": 9, "text": "The chosen model was Quartz-3B."}, {"sent_id": 10, "text": "For evidence retrieval, the 2026-04-23 note says Project Saffron went with Aster-3B."}], "source": "synthetic_rewritten", "title": "Run Brief 145"}
{"doc_id": "doc_000146", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-17."}, {"sent_id": 3, "text": "The chosen model was Cedar-3B."}, {"sent_id": 4, "text": "Project Aster had a failed run on 2026-06-04 involving Orchid-7B and Node Cedar-02."}, {"sent_id": 5, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 6, "text": "The 2026-05-02 compute note links Node Sycamore-13 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 7, "text": "Project Sonata logged a personnel update on 2026-06-14: Sofia Frost took the lead responsibilities."}, {"sent_id": 8, "text": "The 2026-05-14 failure note says Marble-8B on Node Maple-01 failed for Project Saffron because of a checkpoint-mismatch error."}, {"sent_id": 9, "text": "Project Anchor had a method update on 2026-04-26."}, {"sent_id": 10, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}], "source": "synthetic_rewritten", "title": "Review Packet 146"}
{"doc_id": "doc_000147", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-04-07 staffing section, Project Sonata listed Dr. Jonas Rios as the lead."}, {"sent_id": 3, "text": "On 2026-04-08, Project Saffron shifted Milestone R1: 2026-05-02 became 2026-05-06."}, {"sent_id": 4, "text": "Using Node Rowan-09, River-3B produced 0.854 macro F1 on RiverBench for Project Aster on 2026-06-01."}, {"sent_id": 5, "text": "For Project Nereid, the run using Atlas-8B on Node Hazel-14 did not complete on 2026-06-25; the issue was an unstable-validation-loss error."}, {"sent_id": 6, "text": "Project Aster received 3 GPUs from Node Cedar-02 in the 2026-06-05 allocation log."}, {"sent_id": 7, "text": "Using Node Hazel-14, Atlas-8B produced 0.864 macro F1 on RiverBench-2 for Project Nereid on 2026-05-17."}, {"sent_id": 8, "text": "In the 2026-04-28 staffing section, Project Aster listed Mina Adler as the retrieval owner."}, {"sent_id": 9, "text": "Milestone R1 for Project Sonata was revised on 2026-05-26."}, {"sent_id": 10, "text": "The schedule moved from 2026-06-14 to 2026-06-11."}], "source": "synthetic_rewritten", "title": "Run Brief 147"}
{"doc_id": "doc_000148", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Finch-7B was the model selected by Project Anchor on 2026-04-19 for the error analysis track."}, {"sent_id": 3, "text": "On 2026-05-25, the Quiet Retriever Study 1 review reported dense retrieval and said it did not use a reward model."}, {"sent_id": 4, "text": "In the 2026-06-22 staffing section, Project Aster listed Dr. Noah Chen as the retrieval owner."}, {"sent_id": 5, "text": "The 2026-04-12 method entry for Project Sonata moved from threshold search to reward reranking."}, {"sent_id": 6, "text": "The 2026-04-16 failure note says River-7B on Node Rowan-09 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 7, "text": "In the 2026-06-09 staffing section, Project Sonata listed Sofia Frost as the lead."}], "source": "synthetic_rewritten", "title": "Review Packet 148"}
{"doc_id": "doc_000149", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-05-18 staffing section, Project Anchor listed Dr. Arun Kwan as the data steward."}, {"sent_id": 3, "text": "The staffing note for Project Meridian was dated 2026-04-07."}, {"sent_id": 4, "text": "It named Dr. Ravi Nadir for the data steward duties."}, {"sent_id": 5, "text": "On 2026-04-02, Project Meridian retired hard-negative mining in favor of LoRA adaptation."}, {"sent_id": 6, "text": "On 2026-04-02, Project Sonata shifted Milestone R1: 2026-04-11 became 2026-04-16."}, {"sent_id": 7, "text": "Project Anchor logged a personnel update on 2026-04-06: Leo Hale took the data steward responsibilities."}, {"sent_id": 8, "text": "For reranking, the 2026-04-10 note says Project Anchor went with Orchid-8B."}, {"sent_id": 9, "text": "On 2026-05-09, Project Saffron retired hard-negative mining in favor of LoRA adaptation."}], "source": "synthetic_rewritten", "title": "Run Brief 149"}
{"doc_id": "doc_000150", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid received 1 GPU from Node Laurel-11 in the 2026-04-11 allocation log."}, {"sent_id": 3, "text": "Nimbus-7B was the model selected by Project Meridian on 2026-05-27 for the calibration track."}, {"sent_id": 4, "text": "Project Saffron's Milestone X1 deadline change on 2026-04-23 replaced 2026-05-05 with 2026-05-13."}, {"sent_id": 5, "text": "Quartz-7B was the model selected by Project Sonata on 2026-05-01 for the claim classification track."}, {"sent_id": 6, "text": "A compute allocation for Project Saffron was logged on 2026-04-18."}, {"sent_id": 7, "text": "Node Sycamore-13 supplied 2 GPUs."}, {"sent_id": 8, "text": "A compute allocation for Project Anchor was logged on 2026-05-07."}, {"sent_id": 9, "text": "Node Rowan-09 supplied 4 GPUs."}, {"sent_id": 10, "text": "Project Nereid received 3 GPUs from Node Hazel-14 in the 2026-06-27 allocation log."}, {"sent_id": 11, "text": "In the 2026-06-15 staffing section, Project Saffron listed Lena Costa as the data steward."}, {"sent_id": 12, "text": "The 2026-05-22 compute note links Node Cedar-02 to Project Aster with an allocation of 3 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 150"}
{"doc_id": "doc_000151", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-06-30 result for Project Aster put Helix-7B on OrionBench-2 at 0.722 evidence F1, with Node Poplar-12 handling the run."}, {"sent_id": 3, "text": "In the 2026-05-11 staffing section, Project Meridian listed Nadia Chen as the evaluation owner."}, {"sent_id": 4, "text": "Project Meridian had a method update on 2026-06-17."}, {"sent_id": 5, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 6, "text": "On 2026-05-17, Project Anchor retired contrastive tuning in favor of threshold search."}, {"sent_id": 7, "text": "Project Aster logged a personnel update on 2026-06-02: Mina Adler took the retrieval owner responsibilities."}, {"sent_id": 8, "text": "Project Aster had a method update on 2026-05-03."}, {"sent_id": 9, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 10, "text": "Project Anchor had a method update on 2026-04-19."}, {"sent_id": 11, "text": "The old approach was document chunking, and the replacement was metric smoothing."}, {"sent_id": 12, "text": "Milestone D1 for Project Meridian was revised on 2026-04-29."}, {"sent_id": 13, "text": "The schedule moved from 2026-05-21 to 2026-05-27."}], "source": "synthetic_rewritten", "title": "Project Update 151"}
{"doc_id": "doc_000152", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Quiet Retriever Study 2 appeared in a 2026-06-01 review entry using metric smoothing; the note said it did not use a reward model."}, {"sent_id": 3, "text": "Kestrel-7B was the model selected by Project Meridian on 2026-06-02 for the claim classification track."}, {"sent_id": 4, "text": "The 2026-05-17 method entry for Project Aster moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 5, "text": "Project Meridian had a method update on 2026-04-26."}, {"sent_id": 6, "text": "The old approach was reward reranking, and the replacement was data mixing."}, {"sent_id": 7, "text": "On 2026-05-14, Project Aster shifted Milestone H1: 2026-06-02 became 2026-05-30."}, {"sent_id": 8, "text": "Project Anchor received 2 GPUs from Node Juniper-06 in the 2026-06-22 allocation log."}], "source": "synthetic_rewritten", "title": "Lab Digest 152"}
{"doc_id": "doc_000153", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-04-26 method entry for Project Meridian moved from confidence calibration to chain verification."}, {"sent_id": 3, "text": "The 2026-05-23 result for Project Anchor put Finch-3B on VestaLogs at 0.821 latency efficiency score, with Node Juniper-06 handling the run."}, {"sent_id": 4, "text": "For reranking, the 2026-05-13 note says Project Aster went with Orchid-8B."}, {"sent_id": 5, "text": "Project Meridian had a method update on 2026-05-27."}, {"sent_id": 6, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 7, "text": "For error analysis, the 2026-04-26 note says Project Aster went with Finch-8B."}, {"sent_id": 8, "text": "For reranking, the 2026-05-29 note says Project Aster went with Orchid-3B."}], "source": "synthetic_rewritten", "title": "Project Update 153"}
{"doc_id": "doc_000154", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-04-21 staffing section, Project Saffron listed Priya Moss as the data steward."}, {"sent_id": 3, "text": "Project Nereid had a failed run on 2026-04-09 involving Cedar-7B and Node Laurel-11."}, {"sent_id": 4, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 5, "text": "On 2026-04-04, Project Meridian retired cross-encoder reranking in favor of late interaction."}, {"sent_id": 6, "text": "On 2026-05-04, the Quiet Retriever Study 1 review reported evidence pooling and said it did not use a reward model."}, {"sent_id": 7, "text": "For calibration, the 2026-05-10 note says Project Nereid went with Nimbus-7B."}, {"sent_id": 8, "text": "Project Sonata had a failed run on 2026-04-16 involving Aster-3B and Node Willow-05."}, {"sent_id": 9, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 10, "text": "Project Nereid had a failed run on 2026-06-05 involving Kestrel-8B and Node Birch-04."}, {"sent_id": 11, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 12, "text": "For calibration, the 2026-06-03 note says Project Nereid went with Nimbus-8B."}, {"sent_id": 13, "text": "Project Aster had a failed run on 2026-04-16 involving Orchid-7B and Node Cedar-02."}, {"sent_id": 14, "text": "The recorded cause was an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Lab Digest 154"}
{"doc_id": "doc_000155", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian had a failed run on 2026-04-04 involving Atlas-7B and Node Hazel-14."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "For Project Aster, the run using Lumen-8B on Node Spruce-03 did not complete on 2026-04-02; the issue was an unstable-validation-loss error."}, {"sent_id": 5, "text": "Milestone X1 for Project Sonata was revised on 2026-05-28."}, {"sent_id": 6, "text": "The schedule moved from 2026-06-04 to 2026-06-03."}, {"sent_id": 7, "text": "The 2026-05-17 method entry for Project Saffron moved from temporal filtering to alias expansion."}, {"sent_id": 8, "text": "The 2026-05-22 failure note says Mira-3B on Node Sycamore-13 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 9, "text": "The 2026-04-18 method entry for Project Sonata moved from dense retrieval to rank fusion."}, {"sent_id": 10, "text": "Project Nereid recorded a model choice for reranking on 2026-04-22."}, {"sent_id": 11, "text": "The chosen model was Nova-3B."}, {"sent_id": 12, "text": "Using Node Juniper-06, Finch-8B produced 0.851 accuracy on VestaLogs for Project Anchor on 2026-05-30."}, {"sent_id": 13, "text": "Milestone X1 for Project Sonata was revised on 2026-04-14."}, {"sent_id": 14, "text": "The schedule moved from 2026-04-21 to 2026-04-20."}], "source": "synthetic_rewritten", "title": "Project Update 155"}
{"doc_id": "doc_000156", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Milestone L1 for Project Saffron was revised on 2026-05-11."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-27 to 2026-05-31."}, {"sent_id": 4, "text": "Project Aster logged a personnel update on 2026-05-18: Anika Sato took the lead responsibilities."}, {"sent_id": 5, "text": "Project Meridian ran Atlas-8B on SignalSet-3 using Node Hazel-14 on 2026-05-12."}, {"sent_id": 6, "text": "The reported accuracy was 0.759."}, {"sent_id": 7, "text": "On 2026-06-18, Project Aster shifted Milestone B1: 2026-07-09 became 2026-07-18."}, {"sent_id": 8, "text": "For reranking, the 2026-04-05 note says Project Meridian went with Nova-3B."}, {"sent_id": 9, "text": "Project Sonata received 3 GPUs from Node Willow-05 in the 2026-04-04 allocation log."}, {"sent_id": 10, "text": "On 2026-04-29, Project Meridian shifted Milestone V1: 2026-05-15 became 2026-05-19."}], "source": "synthetic_rewritten", "title": "Lab Digest 156"}
{"doc_id": "doc_000157", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-23 failure note says Quartz-7B on Node Pine-07 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 3, "text": "Project Saffron logged a personnel update on 2026-05-11: Dr. Theo Lind took the evaluation owner responsibilities."}, {"sent_id": 4, "text": "For calibration, the 2026-05-14 note says Project Saffron went with Mira-3B."}, {"sent_id": 5, "text": "In the 2026-04-27 staffing section, Project Aster listed Felix Brooks as the retrieval owner."}, {"sent_id": 6, "text": "The 2026-05-23 compute note links Node Aspen-01 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 7, "text": "Project Meridian received 2 GPUs from Node Elm-08 in the 2026-05-16 allocation log."}, {"sent_id": 8, "text": "In the 2026-06-07 staffing section, Project Aster listed Talia Marin as the lead."}, {"sent_id": 9, "text": "The 2026-06-20 compute note links Node Aspen-01 to Project Meridian with an allocation of 2 GPUs."}], "source": "synthetic_rewritten", "title": "Project Update 157"}
{"doc_id": "doc_000158", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Using Node Spruce-03, Lumen-3B produced 0.878 macro F1 on TraceEval for Project Aster on 2026-04-13."}, {"sent_id": 3, "text": "In the 2026-04-13 staffing section, Project Saffron listed Dr. Theo Grant as the data steward."}, {"sent_id": 4, "text": "A compute allocation for Project Aster was logged on 2026-05-15."}, {"sent_id": 5, "text": "Node Cedar-02 supplied 1 GPU."}, {"sent_id": 6, "text": "The 2026-06-11 compute note links Node Hazel-14 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 7, "text": "Project Saffron recorded a model choice for error analysis on 2026-05-05."}, {"sent_id": 8, "text": "The chosen model was Marble-8B."}, {"sent_id": 9, "text": "Quiet Retriever Study 4 appeared in a 2026-04-27 review entry using structured prompting; the note said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Lab Digest 158"}
{"doc_id": "doc_000159", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Vela-7B was the model selected by Project Saffron on 2026-05-27 for the reranking track."}, {"sent_id": 3, "text": "A compute allocation for Project Sonata was logged on 2026-04-18."}, {"sent_id": 4, "text": "Node Pine-07 supplied 1 GPU."}, {"sent_id": 5, "text": "The 2026-06-28 method entry for Project Aster moved from threshold search to reward reranking."}, {"sent_id": 6, "text": "Project Anchor received 2 GPUs from Node Rowan-09 in the 2026-05-08 allocation log."}, {"sent_id": 7, "text": "Project Nereid ran Kestrel-8B on OrionBench-3 using Node Birch-04 on 2026-06-02."}, {"sent_id": 8, "text": "The reported evidence F1 was 0.732."}, {"sent_id": 9, "text": "Quartz-8B was the model selected by Project Saffron on 2026-05-17 for the claim classification track."}, {"sent_id": 10, "text": "The staffing note for Project Anchor was dated 2026-05-04."}, {"sent_id": 11, "text": "It named Dr. Arun Kwan for the data steward duties."}, {"sent_id": 12, "text": "The 2026-06-13 compute note links Node Elm-08 to Project Meridian with an allocation of 2 GPUs."}], "source": "synthetic_rewritten", "title": "Project Update 159"}
{"doc_id": "doc_000160", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-05-17 method entry for Project Meridian moved from structured prompting to QLoRA adaptation."}, {"sent_id": 3, "text": "On 2026-05-15, Project Anchor shifted Milestone N1: 2026-06-06 became 2026-06-16."}, {"sent_id": 4, "text": "River-7B was the model selected by Project Anchor on 2026-04-26 for the evidence retrieval track."}, {"sent_id": 5, "text": "The 2026-04-05 method entry for Project Meridian moved from alias expansion to dense retrieval."}, {"sent_id": 6, "text": "For evidence retrieval, the 2026-05-15 note says Project Anchor went with River-3B."}, {"sent_id": 7, "text": "Project Sonata logged a personnel update on 2026-04-21: Mara Quinn took the lead responsibilities."}], "source": "synthetic_rewritten", "title": "Run Brief 160"}
{"doc_id": "doc_000161", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-06-25, Project Sonata shifted Milestone F1: 2026-07-08 became 2026-07-13."}, {"sent_id": 3, "text": "Project Anchor ran Orchid-7B on LumenFacts using Node Cedar-02 on 2026-06-08."}, {"sent_id": 4, "text": "The reported latency efficiency score was 0.845."}, {"sent_id": 5, "text": "Project Aster's Milestone H1 deadline change on 2026-06-22 replaced 2026-07-11 with 2026-07-08."}, {"sent_id": 6, "text": "Milestone T1 for Project Aster was revised on 2026-05-28."}, {"sent_id": 7, "text": "The schedule moved from 2026-06-22 to 2026-06-27."}, {"sent_id": 8, "text": "A compute allocation for Project Anchor was logged on 2026-04-04."}, {"sent_id": 9, "text": "Node Juniper-06 supplied 2 GPUs."}, {"sent_id": 10, "text": "Nested Verifier Study 4 appeared in a 2026-04-20 review entry using cross-encoder reranking; the note said it did not use a reward model."}, {"sent_id": 11, "text": "The staffing note for Project Anchor was dated 2026-04-14."}, {"sent_id": 12, "text": "It named Selene Kim for the evaluation owner duties."}, {"sent_id": 13, "text": "Project Meridian ran Kestrel-8B on CedarQA using Node Birch-04 on 2026-06-16."}, {"sent_id": 14, "text": "The reported latency efficiency score was 0.597."}, {"sent_id": 15, "text": "The 2026-06-27 result for Project Aster put River-7B on RiverBench at 0.854 macro F1, with Node Rowan-09 handling the run."}], "source": "synthetic_rewritten", "title": "Review Packet 161"}
{"doc_id": "doc_000162", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Nereid's Milestone V1 deadline change on 2026-04-03 replaced 2026-04-24 with 2026-04-29."}, {"sent_id": 3, "text": "The 2026-05-29 failure note says Lumen-7B on Node Spruce-03 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 4, "text": "Project Sonata ran Mira-3B on TraceEval-3 using Node Sycamore-13 on 2026-06-09."}, {"sent_id": 5, "text": "The reported evidence F1 was 0.658."}, {"sent_id": 6, "text": "On 2026-06-11, Project Aster shifted Milestone B1: 2026-06-22 became 2026-06-17."}, {"sent_id": 7, "text": "Project Aster had a failed run on 2026-05-07 involving Orchid-7B and Node Cedar-02."}, {"sent_id": 8, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 9, "text": "A compute allocation for Project Sonata was logged on 2026-04-16."}, {"sent_id": 10, "text": "Node Fir-10 supplied 1 GPU."}, {"sent_id": 11, "text": "The 2026-05-04 result for Project Aster put Orchid-7B on LabQA-3 at 0.890 macro F1, with Node Cedar-02 handling the run."}], "source": "synthetic_rewritten", "title": "Run Brief 162"}
{"doc_id": "doc_000163", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-04-16 method entry for Project Saffron moved from contrastive tuning to threshold search."}, {"sent_id": 3, "text": "On 2026-04-11, Project Sonata retired sentence pruning in favor of cross-encoder reranking."}, {"sent_id": 4, "text": "Project Sonata logged a personnel update on 2026-06-16: Samir Ames took the lead responsibilities."}, {"sent_id": 5, "text": "The 2026-04-24 failure note says Vela-3B on Node Fir-10 failed for Project Saffron because of a missing-index error."}, {"sent_id": 6, "text": "Project Nereid had a method update on 2026-04-05."}, {"sent_id": 7, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 8, "text": "The 2026-05-05 result for Project Sonata put Aster-8B on RiverBench-3 at 0.664 macro F1, with Node Willow-05 handling the run."}], "source": "synthetic_rewritten", "title": "Review Packet 163"}
{"doc_id": "doc_000164", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-10 method entry for Project Saffron moved from cross-encoder reranking to late interaction."}, {"sent_id": 3, "text": "The 2026-05-20 method entry for Project Nereid moved from structured prompting to QLoRA adaptation."}, {"sent_id": 4, "text": "Project Aster received 1 GPU from Node Spruce-03 in the 2026-04-11 allocation log."}, {"sent_id": 5, "text": "The 2026-05-27 method entry for Project Sonata moved from rank fusion to teacher distillation."}, {"sent_id": 6, "text": "Project Aster received 3 GPUs from Node Poplar-12 in the 2026-06-22 allocation log."}, {"sent_id": 7, "text": "The staffing note for Project Nereid was dated 2026-04-20."}, {"sent_id": 8, "text": "It named Vera Kim for the lead duties."}, {"sent_id": 9, "text": "Project Sonata ran Quartz-8B on OrionBench using Node Pine-07 on 2026-04-06."}, {"sent_id": 10, "text": "The reported macro F1 was 0.562."}], "source": "synthetic_rewritten", "title": "Run Brief 164"}
{"doc_id": "doc_000165", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-06-14, Project Nereid retired teacher distillation in favor of query rewriting."}, {"sent_id": 3, "text": "Using Node Pine-07, Quartz-8B produced 0.847 accuracy on CedarQA-2 for Project Saffron on 2026-04-14."}, {"sent_id": 4, "text": "The 2026-05-28 failure note says Quartz-8B on Node Pine-07 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 5, "text": "The 2026-04-11 compute note links Node Sycamore-13 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 6, "text": "For error analysis, the 2026-04-23 note says Project Saffron went with Marble-7B."}, {"sent_id": 7, "text": "Nova-8B was the model selected by Project Meridian on 2026-05-10 for the reranking track."}], "source": "synthetic_rewritten", "title": "Review Packet 165"}
{"doc_id": "doc_000166", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata had a method update on 2026-04-29."}, {"sent_id": 3, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 4, "text": "Using Node Juniper-06, Finch-8B produced 0.611 latency efficiency score on VestaLogs for Project Anchor on 2026-06-21."}, {"sent_id": 5, "text": "On 2026-05-23, Project Sonata retired threshold search in favor of reward reranking."}, {"sent_id": 6, "text": "The staffing note for Project Sonata was dated 2026-04-20."}, {"sent_id": 7, "text": "It named Mara Lane for the retrieval owner duties."}, {"sent_id": 8, "text": "On 2026-05-19, Project Sonata shifted Milestone R1: 2026-05-28 became 2026-06-02."}, {"sent_id": 9, "text": "The review note for Amber Ranking Study 4 was dated 2026-06-08."}, {"sent_id": 10, "text": "It described the method as LoRA adaptation and said it did not use a reward model."}, {"sent_id": 11, "text": "The 2026-06-27 method entry for Project Anchor moved from dense retrieval to rank fusion."}, {"sent_id": 12, "text": "The 2026-04-12 method entry for Project Meridian moved from teacher distillation to query rewriting."}], "source": "synthetic_rewritten", "title": "Run Brief 166"}
{"doc_id": "doc_000167", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-12, Project Sonata retired hybrid retrieval in favor of document chunking."}, {"sent_id": 3, "text": "Project Meridian had a method update on 2026-06-18."}, {"sent_id": 4, "text": "The old approach was threshold search, and the replacement was reward reranking."}, {"sent_id": 5, "text": "For error analysis, the 2026-04-02 note says Project Aster went with Finch-7B."}, {"sent_id": 6, "text": "A compute allocation for Project Nereid was logged on 2026-04-24."}, {"sent_id": 7, "text": "Node Hazel-14 supplied 3 GPUs."}, {"sent_id": 8, "text": "Project Aster ran Helix-8B on OrionBench-2 using Node Poplar-12 on 2026-06-13."}, {"sent_id": 9, "text": "The reported evidence F1 was 0.872."}, {"sent_id": 10, "text": "In the 2026-04-14 staffing section, Project Anchor listed Leo Hale as the data steward."}, {"sent_id": 11, "text": "Project Saffron's Milestone D2 deadline change on 2026-04-22 replaced 2026-05-02 with 2026-05-12."}, {"sent_id": 12, "text": "Using Node Cedar-02, Orchid-8B produced 0.845 accuracy on LumenFacts for Project Anchor on 2026-05-26."}, {"sent_id": 13, "text": "Project Meridian ran Atlas-7B on SignalSet-3 using Node Hazel-14 on 2026-05-19."}, {"sent_id": 14, "text": "The reported latency efficiency score was 0.789."}], "source": "synthetic_rewritten", "title": "Review Packet 167"}
{"doc_id": "doc_000168", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-03 method entry for Project Anchor moved from late interaction to BM25 retrieval."}, {"sent_id": 3, "text": "On 2026-04-19, Project Nereid retired confidence calibration in favor of chain verification."}, {"sent_id": 4, "text": "For Project Anchor, the run using Helix-7B on Node Poplar-12 did not complete on 2026-06-25; the issue was a missing-index error."}, {"sent_id": 5, "text": "Project Aster had a method update on 2026-04-23."}, {"sent_id": 6, "text": "The old approach was hard-negative mining, and the replacement was LoRA adaptation."}, {"sent_id": 7, "text": "The 2026-04-11 compute note links Node Willow-05 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 8, "text": "Project Meridian had a method update on 2026-06-03."}, {"sent_id": 9, "text": "The old approach was cross-encoder reranking, and the replacement was late interaction."}, {"sent_id": 10, "text": "Marble-7B was the model selected by Project Sonata on 2026-06-12 for the error analysis track."}, {"sent_id": 11, "text": "The 2026-05-08 failure note says Vela-7B on Node Fir-10 failed for Project Saffron because of a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 168"}
{"doc_id": "doc_000169", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid ran Kestrel-7B on OrionBench-3 using Node Birch-04 on 2026-06-09."}, {"sent_id": 3, "text": "The reported macro F1 was 0.762."}, {"sent_id": 4, "text": "Using Node Rowan-09, River-7B produced 0.854 evidence F1 on RiverBench for Project Aster on 2026-05-19."}, {"sent_id": 5, "text": "Aster-8B was the model selected by Project Sonata on 2026-04-08 for the evidence retrieval track."}, {"sent_id": 6, "text": "The 2026-06-02 result for Project Anchor put Lumen-3B on MemoTrace-3 at 0.563 accuracy, with Node Spruce-03 handling the run."}, {"sent_id": 7, "text": "In the 2026-05-05 staffing section, Project Nereid listed Evan Moss as the lead."}, {"sent_id": 8, "text": "Project Anchor had a failed run on 2026-04-25 involving Orchid-3B and Node Cedar-02."}, {"sent_id": 9, "text": "The recorded cause was a missing-index error."}, {"sent_id": 10, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-04-17."}, {"sent_id": 11, "text": "The chosen model was Atlas-7B."}, {"sent_id": 12, "text": "In the 2026-06-30 staffing section, Project Aster listed Anika Sato as the lead."}], "source": "synthetic_rewritten", "title": "Review Packet 169"}
{"doc_id": "doc_000170", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Milestone Z1 for Project Aster was revised on 2026-06-02."}, {"sent_id": 3, "text": "The schedule moved from 2026-06-15 to 2026-06-20."}, {"sent_id": 4, "text": "Project Saffron had a method update on 2026-04-12."}, {"sent_id": 5, "text": "The old approach was reward reranking, and the replacement was data mixing."}, {"sent_id": 6, "text": "The 2026-04-09 compute note links Node Birch-04 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 7, "text": "The 2026-05-26 result for Project Anchor put River-8B on SignalSet-2 at 0.689 accuracy, with Node Rowan-09 handling the run."}, {"sent_id": 8, "text": "Project Saffron had a method update on 2026-05-21."}, {"sent_id": 9, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 10, "text": "The 2026-06-20 compute note links Node Birch-04 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 11, "text": "Project Saffron ran Quartz-7B on CedarQA-2 using Node Pine-07 on 2026-04-12."}, {"sent_id": 12, "text": "The reported accuracy was 0.727."}, {"sent_id": 13, "text": "The 2026-06-20 method entry for Project Aster moved from query rewriting to sentence pruning."}], "source": "synthetic_rewritten", "title": "Project Update 170"}
{"doc_id": "doc_000171", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron recorded a model choice for claim classification on 2026-05-31."}, {"sent_id": 3, "text": "The chosen model was Quartz-3B."}, {"sent_id": 4, "text": "The staffing note for Project Saffron was dated 2026-04-12."}, {"sent_id": 5, "text": "It named Dr. Theo Grant for the data steward duties."}, {"sent_id": 6, "text": "Project Anchor had a method update on 2026-05-10."}, {"sent_id": 7, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 8, "text": "Using Node Sycamore-13, Mira-3B produced 0.613 accuracy on MemoTrace-2 for Project Saffron on 2026-04-14."}, {"sent_id": 9, "text": "The 2026-04-05 result for Project Sonata put Marble-3B on NereidNotes at 0.736 evidence F1, with Node Maple-01 handling the run."}, {"sent_id": 10, "text": "The 2026-05-30 compute note links Node Willow-05 to Project Saffron with an allocation of 4 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 171"}
{"doc_id": "doc_000172", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "A compute allocation for Project Anchor was logged on 2026-05-15."}, {"sent_id": 3, "text": "Node Poplar-12 supplied 2 GPUs."}, {"sent_id": 4, "text": "A compute allocation for Project Aster was logged on 2026-04-18."}, {"sent_id": 5, "text": "Node Spruce-03 supplied 3 GPUs."}, {"sent_id": 6, "text": "The 2026-04-09 method entry for Project Sonata moved from late interaction to BM25 retrieval."}, {"sent_id": 7, "text": "The 2026-05-04 result for Project Meridian put Kestrel-8B on CedarQA at 0.747 latency efficiency score, with Node Birch-04 handling the run."}, {"sent_id": 8, "text": "Kestrel-7B was the model selected by Project Nereid on 2026-04-19 for the claim classification track."}, {"sent_id": 9, "text": "For calibration, the 2026-04-26 note says Project Saffron went with Mira-3B."}, {"sent_id": 10, "text": "Project Sonata had a method update on 2026-06-07."}, {"sent_id": 11, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}], "source": "synthetic_rewritten", "title": "Project Update 172"}
{"doc_id": "doc_000173", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-05-20, Project Nereid shifted Milestone J1: 2026-06-14 became 2026-06-23."}, {"sent_id": 3, "text": "The 2026-06-10 method entry for Project Sonata moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 4, "text": "On 2026-06-26, Project Saffron shifted Milestone L1: 2026-07-12 became 2026-07-20."}, {"sent_id": 5, "text": "The 2026-04-25 compute note links Node Poplar-12 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 6, "text": "The 2026-06-04 failure note says Aster-7B on Node Willow-05 failed for Project Saffron because of a checkpoint-mismatch error."}, {"sent_id": 7, "text": "The 2026-04-22 method entry for Project Aster moved from sentence pruning to cross-encoder reranking."}], "source": "synthetic_rewritten", "title": "Lab Digest 173"}
{"doc_id": "doc_000174", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-16 result for Project Sonata put Aster-3B on RiverBench-3 at 0.544 evidence F1, with Node Willow-05 handling the run."}, {"sent_id": 3, "text": "Project Anchor had a method update on 2026-04-09."}, {"sent_id": 4, "text": "The old approach was metric smoothing, and the replacement was hard-negative mining."}, {"sent_id": 5, "text": "Project Meridian logged a personnel update on 2026-05-05: Elian Shah took the data steward responsibilities."}, {"sent_id": 6, "text": "Project Anchor received 4 GPUs from Node Rowan-09 in the 2026-05-29 allocation log."}, {"sent_id": 7, "text": "Milestone P1 for Project Nereid was revised on 2026-04-08."}, {"sent_id": 8, "text": "The schedule moved from 2026-04-21 to 2026-04-30."}, {"sent_id": 9, "text": "The staffing note for Project Sonata was dated 2026-05-05."}, {"sent_id": 10, "text": "It named Rohan Park for the lead duties."}, {"sent_id": 11, "text": "The 2026-06-25 failure note says Atlas-7B on Node Hazel-14 failed for Project Meridian because of a missing-index error."}, {"sent_id": 12, "text": "On 2026-05-29, Project Saffron shifted Milestone R1: 2026-06-22 became 2026-06-26."}], "source": "synthetic_rewritten", "title": "Project Update 174"}
{"doc_id": "doc_000175", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Nereid had a failed run on 2026-06-11 involving Kestrel-8B and Node Birch-04."}, {"sent_id": 3, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 4, "text": "Using Node Rowan-09, River-8B produced 0.824 evidence F1 on RiverBench for Project Aster on 2026-06-20."}, {"sent_id": 5, "text": "Project Saffron had a method update on 2026-04-15."}, {"sent_id": 6, "text": "The old approach was cross-encoder reranking, and the replacement was late interaction."}, {"sent_id": 7, "text": "On 2026-05-16, Project Nereid retired rank fusion in favor of teacher distillation."}, {"sent_id": 8, "text": "On 2026-04-25, Project Sonata retired metric smoothing in favor of hard-negative mining."}, {"sent_id": 9, "text": "Project Meridian had a failed run on 2026-05-08 involving Kestrel-7B and Node Birch-04."}, {"sent_id": 10, "text": "The recorded cause was a missing-index error."}, {"sent_id": 11, "text": "The 2026-06-28 method entry for Project Aster moved from structured prompting to QLoRA adaptation."}, {"sent_id": 12, "text": "Project Nereid recorded a model choice for reranking on 2026-05-10."}, {"sent_id": 13, "text": "The chosen model was Nova-3B."}, {"sent_id": 14, "text": "Project Meridian ran Cedar-8B on VestaLogs-2 using Node Laurel-11 on 2026-06-09."}, {"sent_id": 15, "text": "The reported latency efficiency score was 0.801."}], "source": "synthetic_rewritten", "title": "Lab Digest 175"}
{"doc_id": "doc_000176", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-27 compute note links Node Willow-05 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "Project Nereid received 3 GPUs from Node Aspen-01 in the 2026-06-18 allocation log."}, {"sent_id": 4, "text": "Milestone P1 for Project Nereid was revised on 2026-05-21."}, {"sent_id": 5, "text": "The schedule moved from 2026-06-03 to 2026-06-12."}, {"sent_id": 6, "text": "Project Saffron logged a personnel update on 2026-05-04: Priya Vale took the evaluation owner responsibilities."}, {"sent_id": 7, "text": "Project Sonata's Milestone R1 deadline change on 2026-05-20 replaced 2026-05-29 with 2026-06-03."}, {"sent_id": 8, "text": "Milestone B1 for Project Anchor was revised on 2026-06-09."}, {"sent_id": 9, "text": "The schedule moved from 2026-06-25 to 2026-06-29."}, {"sent_id": 10, "text": "Project Aster had a method update on 2026-04-05."}, {"sent_id": 11, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}], "source": "synthetic_rewritten", "title": "Project Update 176"}
{"doc_id": "doc_000177", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-05-06, Project Meridian retired document chunking in favor of metric smoothing."}, {"sent_id": 3, "text": "Project Nereid ran Cedar-7B on NereidNotes-3 using Node Laurel-11 on 2026-05-09."}, {"sent_id": 4, "text": "The reported macro F1 was 0.696."}, {"sent_id": 5, "text": "Mira-8B was the model selected by Project Sonata on 2026-06-25 for the calibration track."}, {"sent_id": 6, "text": "For evidence retrieval, the 2026-06-26 note says Project Aster went with River-8B."}, {"sent_id": 7, "text": "For error analysis, the 2026-04-09 note says Project Meridian went with Cedar-8B."}, {"sent_id": 8, "text": "The 2026-06-22 result for Project Meridian put Nimbus-8B on MemoTrace at 0.723 latency efficiency score, with Node Elm-08 handling the run."}, {"sent_id": 9, "text": "Project Saffron had a method update on 2026-05-06."}, {"sent_id": 10, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 11, "text": "The staffing note for Project Nereid was dated 2026-06-29."}, {"sent_id": 12, "text": "It named Evan Iyer for the retrieval owner duties."}, {"sent_id": 13, "text": "For claim classification, the 2026-05-22 note says Project Meridian went with Kestrel-8B."}], "source": "synthetic_rewritten", "title": "Lab Digest 177"}
{"doc_id": "doc_000178", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Anchor recorded a model choice for evidence retrieval on 2026-05-13."}, {"sent_id": 3, "text": "The chosen model was River-3B."}, {"sent_id": 4, "text": "In the 2026-05-12 staffing section, Project Meridian listed Nadia Singh as the data steward."}, {"sent_id": 5, "text": "Project Aster ran Helix-7B on OrionBench-2 using Node Poplar-12 on 2026-06-20."}, {"sent_id": 6, "text": "The reported macro F1 was 0.512."}, {"sent_id": 7, "text": "The staffing note for Project Meridian was dated 2026-06-15."}, {"sent_id": 8, "text": "It named Elian Shah for the data steward duties."}, {"sent_id": 9, "text": "In the 2026-04-13 staffing section, Project Meridian listed Dr. Ravi Nadir as the data steward."}, {"sent_id": 10, "text": "The 2026-05-30 method entry for Project Meridian moved from query rewriting to sentence pruning."}, {"sent_id": 11, "text": "Project Anchor had a method update on 2026-06-13."}, {"sent_id": 12, "text": "The old approach was late interaction, and the replacement was BM25 retrieval."}, {"sent_id": 13, "text": "For claim classification, the 2026-06-05 note says Project Meridian went with Kestrel-3B."}, {"sent_id": 14, "text": "In the 2026-05-05 staffing section, Project Anchor listed Dr. Arun Bauer as the evaluation owner."}], "source": "synthetic_rewritten", "title": "Project Update 178"}
{"doc_id": "doc_000179", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The staffing note for Project Anchor was dated 2026-04-06."}, {"sent_id": 3, "text": "It named Selene Kim for the evaluation owner duties."}, {"sent_id": 4, "text": "The 2026-04-04 result for Project Sonata put Mira-8B on TraceEval-3 at 0.598 macro F1, with Node Sycamore-13 handling the run."}, {"sent_id": 5, "text": "On 2026-06-14, Project Saffron retired BM25 retrieval in favor of evidence pooling."}, {"sent_id": 6, "text": "Project Nereid received 3 GPUs from Node Elm-08 in the 2026-04-11 allocation log."}, {"sent_id": 7, "text": "For claim classification, the 2026-05-03 note says Project Anchor went with Helix-3B."}, {"sent_id": 8, "text": "Project Nereid had a failed run on 2026-06-18 involving Nimbus-8B and Node Elm-08."}, {"sent_id": 9, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 10, "text": "Project Saffron had a failed run on 2026-06-26 involving Aster-3B and Node Willow-05."}, {"sent_id": 11, "text": "The recorded cause was a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Lab Digest 179"}
{"doc_id": "doc_000180", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Aspen-01, Nova-7B produced 0.765 latency efficiency score on LumenFacts-2 for Project Meridian on 2026-04-07."}, {"sent_id": 3, "text": "For evidence retrieval, the 2026-04-09 note says Project Saffron went with Aster-8B."}, {"sent_id": 4, "text": "Project Saffron logged a personnel update on 2026-04-21: Petra Gray took the evaluation owner responsibilities."}, {"sent_id": 5, "text": "For Project Anchor, the run using River-7B on Node Rowan-09 did not complete on 2026-05-30; the issue was a checkpoint-mismatch error."}, {"sent_id": 6, "text": "The 2026-05-07 failure note says Kestrel-8B on Node Birch-04 failed for Project Meridian because of a checkpoint-mismatch error."}, {"sent_id": 7, "text": "On 2026-06-24, Project Anchor shifted Milestone B1: 2026-07-10 became 2026-07-18."}, {"sent_id": 8, "text": "The 2026-04-18 method entry for Project Anchor moved from evidence pooling to calibrated voting."}], "source": "synthetic_rewritten", "title": "Review Packet 180"}
{"doc_id": "doc_000181", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron ran Marble-7B on VestaLogs-3 using Node Maple-01 on 2026-04-14."}, {"sent_id": 3, "text": "The reported latency efficiency score was 0.691."}, {"sent_id": 4, "text": "On 2026-06-12, Project Nereid shifted Milestone J1: 2026-07-07 became 2026-07-16."}, {"sent_id": 5, "text": "Project Nereid had a failed run on 2026-04-10 involving Atlas-8B and Node Hazel-14."}, {"sent_id": 6, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 7, "text": "Project Aster logged a personnel update on 2026-04-20: Felix Lind took the lead responsibilities."}, {"sent_id": 8, "text": "For calibration, the 2026-05-29 note says Project Nereid went with Nimbus-3B."}, {"sent_id": 9, "text": "The 2026-04-11 compute note links Node Maple-01 to Project Saffron with an allocation of 2 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 181"}
{"doc_id": "doc_000182", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Helix-7B was the model selected by Project Aster on 2026-05-24 for the claim classification track."}, {"sent_id": 3, "text": "On 2026-05-16, Project Meridian retired metric smoothing in favor of hard-negative mining."}, {"sent_id": 4, "text": "Milestone V1 for Project Nereid was revised on 2026-06-02."}, {"sent_id": 5, "text": "The schedule moved from 2026-06-13 to 2026-06-08."}, {"sent_id": 6, "text": "Milestone B2 for Project Meridian was revised on 2026-04-10."}, {"sent_id": 7, "text": "The schedule moved from 2026-05-04 to 2026-05-08."}, {"sent_id": 8, "text": "Project Sonata received 3 GPUs from Node Maple-01 in the 2026-06-05 allocation log."}, {"sent_id": 9, "text": "The 2026-05-19 result for Project Sonata put Marble-7B on NereidNotes at 0.646 evidence F1, with Node Maple-01 handling the run."}], "source": "synthetic_rewritten", "title": "Review Packet 182"}
{"doc_id": "doc_000183", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata recorded a model choice for calibration on 2026-06-07."}, {"sent_id": 3, "text": "The chosen model was Mira-8B."}, {"sent_id": 4, "text": "The staffing note for Project Sonata was dated 2026-05-19."}, {"sent_id": 5, "text": "It named Dr. Jonas Rios for the lead duties."}, {"sent_id": 6, "text": "A compute allocation for Project Anchor was logged on 2026-05-09."}, {"sent_id": 7, "text": "Node Cedar-02 supplied 4 GPUs."}, {"sent_id": 8, "text": "Milestone R1 for Project Saffron was revised on 2026-05-08."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-22 to 2026-06-01."}, {"sent_id": 10, "text": "Project Saffron recorded a model choice for reranking on 2026-04-26."}, {"sent_id": 11, "text": "The chosen model was Vela-7B."}, {"sent_id": 12, "text": "Project Anchor received 4 GPUs from Node Spruce-03 in the 2026-06-18 allocation log."}, {"sent_id": 13, "text": "Project Saffron had a method update on 2026-05-16."}, {"sent_id": 14, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 15, "text": "Project Nereid received 1 GPU from Node Hazel-14 in the 2026-04-17 allocation log."}], "source": "synthetic_rewritten", "title": "Run Brief 183"}
{"doc_id": "doc_000184", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for calibration on 2026-06-07."}, {"sent_id": 3, "text": "The chosen model was Nimbus-7B."}, {"sent_id": 4, "text": "A compute allocation for Project Nereid was logged on 2026-05-09."}, {"sent_id": 5, "text": "Node Birch-04 supplied 3 GPUs."}, {"sent_id": 6, "text": "Project Meridian received 2 GPUs from Node Laurel-11 in the 2026-06-25 allocation log."}, {"sent_id": 7, "text": "Milestone V1 for Project Nereid was revised on 2026-04-17."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-08 to 2026-05-17."}, {"sent_id": 9, "text": "The review note for Quiet Retriever Study 4 was dated 2026-06-03."}, {"sent_id": 10, "text": "It described the method as structured prompting and said it did not use a reward model."}, {"sent_id": 11, "text": "The staffing note for Project Anchor was dated 2026-04-28."}, {"sent_id": 12, "text": "It named Kira Frost for the data steward duties."}, {"sent_id": 13, "text": "A compute allocation for Project Aster was logged on 2026-05-16."}, {"sent_id": 14, "text": "Node Juniper-06 supplied 3 GPUs."}, {"sent_id": 15, "text": "The 2026-04-30 failure note says Helix-7B on Node Poplar-12 failed for Project Aster because of an unstable-validation-loss error."}, {"sent_id": 16, "text": "Project Meridian logged a personnel update on 2026-05-18: Mira Sato took the data steward responsibilities."}], "source": "synthetic_rewritten", "title": "Review Packet 184"}
{"doc_id": "doc_000185", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-04-06 result for Project Nereid put Cedar-7B on NereidNotes-3 at 0.666 evidence F1, with Node Laurel-11 handling the run."}, {"sent_id": 3, "text": "Project Meridian recorded a model choice for reranking on 2026-05-12."}, {"sent_id": 4, "text": "The chosen model was Nova-3B."}, {"sent_id": 5, "text": "Project Nereid received 1 GPU from Node Birch-04 in the 2026-05-30 allocation log."}, {"sent_id": 6, "text": "The staffing note for Project Nereid was dated 2026-06-22."}, {"sent_id": 7, "text": "It named Nora Sol for the retrieval owner duties."}, {"sent_id": 8, "text": "Project Sonata recorded a model choice for calibration on 2026-04-05."}, {"sent_id": 9, "text": "The chosen model was Mira-8B."}, {"sent_id": 10, "text": "On 2026-04-07, Project Saffron shifted Milestone R1: 2026-05-01 became 2026-05-05."}, {"sent_id": 11, "text": "The 2026-04-03 failure note says River-3B on Node Rowan-09 failed for Project Anchor because of a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Run Brief 185"}
{"doc_id": "doc_000186", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Aster had a failed run on 2026-06-25 involving Helix-3B and Node Poplar-12."}, {"sent_id": 3, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 4, "text": "Milestone V1 for Project Nereid was revised on 2026-04-02."}, {"sent_id": 5, "text": "The schedule moved from 2026-04-13 to 2026-04-08."}, {"sent_id": 6, "text": "For reranking, the 2026-06-17 note says Project Aster went with Orchid-3B."}, {"sent_id": 7, "text": "Project Sonata received 1 GPU from Node Willow-05 in the 2026-05-09 allocation log."}, {"sent_id": 8, "text": "Milestone X1 for Project Sonata was revised on 2026-04-21."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-08 to 2026-05-17."}, {"sent_id": 10, "text": "The 2026-06-26 failure note says Vela-8B on Node Fir-10 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 11, "text": "Project Meridian had a method update on 2026-04-18."}, {"sent_id": 12, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 13, "text": "On 2026-04-05, Project Saffron retired hybrid retrieval in favor of document chunking."}], "source": "synthetic_rewritten", "title": "Review Packet 186"}
{"doc_id": "doc_000187", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-11 result for Project Meridian put Cedar-3B on VestaLogs-2 at 0.771 latency efficiency score, with Node Laurel-11 handling the run."}, {"sent_id": 3, "text": "The 2026-05-27 method entry for Project Meridian moved from hard-negative mining to LoRA adaptation."}, {"sent_id": 4, "text": "Nimbus-8B was the model selected by Project Nereid on 2026-06-09 for the calibration track."}, {"sent_id": 5, "text": "The staffing note for Project Saffron was dated 2026-06-30."}, {"sent_id": 6, "text": "It named Dr. Theo Lind for the evaluation owner duties."}, {"sent_id": 7, "text": "In the 2026-04-27 staffing section, Project Meridian listed Dr. Ravi Nadir as the data steward."}, {"sent_id": 8, "text": "On 2026-06-19, Project Aster shifted Milestone B1: 2026-06-30 became 2026-06-25."}, {"sent_id": 9, "text": "Project Anchor had a method update on 2026-05-24."}, {"sent_id": 10, "text": "The old approach was reward reranking, and the replacement was data mixing."}, {"sent_id": 11, "text": "On 2026-04-25, Project Anchor retired document chunking in favor of metric smoothing."}], "source": "synthetic_rewritten", "title": "Run Brief 187"}
{"doc_id": "doc_000188", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone T1 for Project Aster was revised on 2026-05-21."}, {"sent_id": 3, "text": "The schedule moved from 2026-06-05 to 2026-06-01."}, {"sent_id": 4, "text": "The 2026-05-08 compute note links Node Sycamore-13 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 5, "text": "Milestone F1 for Project Sonata was revised on 2026-05-20."}, {"sent_id": 6, "text": "The schedule moved from 2026-06-12 to 2026-06-10."}, {"sent_id": 7, "text": "A compute allocation for Project Anchor was logged on 2026-05-27."}, {"sent_id": 8, "text": "Node Rowan-09 supplied 4 GPUs."}, {"sent_id": 9, "text": "Milestone R1 for Project Saffron was revised on 2026-05-28."}, {"sent_id": 10, "text": "The schedule moved from 2026-06-11 to 2026-06-21."}, {"sent_id": 11, "text": "Using Node Hazel-14, Atlas-3B produced 0.729 accuracy on SignalSet-3 for Project Meridian on 2026-06-13."}, {"sent_id": 12, "text": "Project Aster recorded a model choice for evidence retrieval on 2026-06-19."}, {"sent_id": 13, "text": "The chosen model was River-3B."}], "source": "synthetic_rewritten", "title": "Review Packet 188"}
{"doc_id": "doc_000189", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron ran Aster-8B on SignalSet using Node Willow-05 on 2026-06-15."}, {"sent_id": 3, "text": "The reported latency efficiency score was 0.589."}, {"sent_id": 4, "text": "The 2026-06-14 result for Project Meridian put Kestrel-7B on CedarQA at 0.867 latency efficiency score, with Node Birch-04 handling the run."}, {"sent_id": 5, "text": "Project Nereid ran Nimbus-7B on TraceEval-2 using Node Elm-08 on 2026-04-28."}, {"sent_id": 6, "text": "The reported macro F1 was 0.738."}, {"sent_id": 7, "text": "Project Meridian ran Cedar-3B on VestaLogs-2 using Node Laurel-11 on 2026-05-18."}, {"sent_id": 8, "text": "The reported latency efficiency score was 0.651."}, {"sent_id": 9, "text": "The 2026-05-18 result for Project Sonata put Marble-8B on NereidNotes at 0.586 macro F1, with Node Maple-01 handling the run."}, {"sent_id": 10, "text": "Project Nereid recorded a model choice for reranking on 2026-06-12."}, {"sent_id": 11, "text": "The chosen model was Nova-3B."}, {"sent_id": 12, "text": "Using Node Birch-04, Kestrel-7B produced 0.807 accuracy on CedarQA for Project Meridian on 2026-05-05."}, {"sent_id": 13, "text": "Project Saffron ran Marble-7B on VestaLogs-3 using Node Maple-01 on 2026-05-02."}, {"sent_id": 14, "text": "The reported latency efficiency score was 0.601."}, {"sent_id": 15, "text": "The 2026-04-02 failure note says River-7B on Node Rowan-09 failed for Project Anchor because of a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 189"}
{"doc_id": "doc_000190", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-27 compute note links Node Fir-10 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 3, "text": "Using Node Maple-01, Marble-7B produced 0.526 macro F1 on NereidNotes for Project Sonata on 2026-05-04."}, {"sent_id": 4, "text": "In the 2026-04-21 staffing section, Project Meridian listed Camila Quinn as the evaluation owner."}, {"sent_id": 5, "text": "The 2026-05-04 compute note links Node Elm-08 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 6, "text": "On 2026-06-29, Project Sonata shifted Milestone L1: 2026-07-10 became 2026-07-05."}, {"sent_id": 7, "text": "Project Nereid logged a personnel update on 2026-05-25: Vera Torres took the retrieval owner responsibilities."}, {"sent_id": 8, "text": "The 2026-05-26 result for Project Aster put River-3B on RiverBench at 0.884 macro F1, with Node Rowan-09 handling the run."}, {"sent_id": 9, "text": "Project Saffron's Milestone X1 deadline change on 2026-04-10 replaced 2026-05-02 with 2026-05-08."}], "source": "synthetic_rewritten", "title": "Lab Digest 190"}
{"doc_id": "doc_000191", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron received 2 GPUs from Node Sycamore-13 in the 2026-06-08 allocation log."}, {"sent_id": 3, "text": "The staffing note for Project Anchor was dated 2026-06-01."}, {"sent_id": 4, "text": "It named Iris Stone for the data steward duties."}, {"sent_id": 5, "text": "In the 2026-06-15 staffing section, Project Saffron listed Owen Marin as the data steward."}, {"sent_id": 6, "text": "Project Sonata recorded a model choice for reranking on 2026-04-24."}, {"sent_id": 7, "text": "The chosen model was Vela-7B."}, {"sent_id": 8, "text": "On 2026-06-06, Project Meridian retired document chunking in favor of metric smoothing."}, {"sent_id": 9, "text": "Project Nereid had a method update on 2026-04-22."}, {"sent_id": 10, "text": "The old approach was calibrated voting, and the replacement was confidence calibration."}, {"sent_id": 11, "text": "Project Saffron ran Quartz-7B on CedarQA-2 using Node Pine-07 on 2026-05-12."}, {"sent_id": 12, "text": "The reported accuracy was 0.577."}], "source": "synthetic_rewritten", "title": "Project Update 191"}
{"doc_id": "doc_000192", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-06-17, Project Aster shifted Milestone H1: 2026-07-06 became 2026-07-03."}, {"sent_id": 3, "text": "Project Nereid ran Cedar-3B on NereidNotes-3 using Node Laurel-11 on 2026-04-13."}, {"sent_id": 4, "text": "The reported macro F1 was 0.696."}, {"sent_id": 5, "text": "The 2026-05-13 method entry for Project Saffron moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 6, "text": "For Project Nereid, the run using Atlas-3B on Node Hazel-14 did not complete on 2026-04-03; the issue was an unstable-validation-loss error."}, {"sent_id": 7, "text": "Atlas-7B was the model selected by Project Meridian on 2026-06-18 for the evidence retrieval track."}, {"sent_id": 8, "text": "Project Saffron recorded a model choice for calibration on 2026-04-01."}, {"sent_id": 9, "text": "The chosen model was Mira-7B."}, {"sent_id": 10, "text": "The 2026-05-23 result for Project Sonata put Aster-8B on RiverBench-3 at 0.574 macro F1, with Node Willow-05 handling the run."}, {"sent_id": 11, "text": "Project Meridian logged a personnel update on 2026-04-07: Mira Nolan took the evaluation owner responsibilities."}, {"sent_id": 12, "text": "A compute allocation for Project Anchor was logged on 2026-04-04."}, {"sent_id": 13, "text": "Node Rowan-09 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 192"}
{"doc_id": "doc_000193", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Aster received 3 GPUs from Node Rowan-09 in the 2026-06-06 allocation log."}, {"sent_id": 3, "text": "The 2026-05-29 compute note links Node Fir-10 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 4, "text": "Project Meridian had a failed run on 2026-04-02 involving Nimbus-7B and Node Elm-08."}, {"sent_id": 5, "text": "The recorded cause was a missing-index error."}, {"sent_id": 6, "text": "The 2026-05-22 compute note links Node Willow-05 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 7, "text": "On 2026-06-14, Project Saffron retired dense retrieval in favor of rank fusion."}, {"sent_id": 8, "text": "Project Nereid's Milestone V1 deadline change on 2026-06-01 replaced 2026-06-22 with 2026-06-27."}, {"sent_id": 9, "text": "In the 2026-04-20 staffing section, Project Saffron listed Owen Marin as the data steward."}], "source": "synthetic_rewritten", "title": "Project Update 193"}
{"doc_id": "doc_000194", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Anchor received 4 GPUs from Node Spruce-03 in the 2026-05-02 allocation log."}, {"sent_id": 3, "text": "Milestone F1 for Project Saffron was revised on 2026-06-15."}, {"sent_id": 4, "text": "The schedule moved from 2026-07-03 to 2026-07-09."}, {"sent_id": 5, "text": "Milestone D1 for Project Nereid was revised on 2026-06-15."}, {"sent_id": 6, "text": "The schedule moved from 2026-06-22 to 2026-06-21."}, {"sent_id": 7, "text": "Kestrel-8B was the model selected by Project Nereid on 2026-04-02 for the claim classification track."}, {"sent_id": 8, "text": "On 2026-05-16, Project Anchor retired data mixing in favor of contrastive tuning."}, {"sent_id": 9, "text": "Using Node Maple-01, Marble-7B produced 0.751 accuracy on VestaLogs-3 for Project Saffron on 2026-05-24."}, {"sent_id": 10, "text": "Marble-3B was the model selected by Project Saffron on 2026-04-30 for the error analysis track."}, {"sent_id": 11, "text": "The 2026-04-08 compute note links Node Spruce-03 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 12, "text": "A compute allocation for Project Aster was logged on 2026-05-08."}, {"sent_id": 13, "text": "Node Juniper-06 supplied 3 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 194"}
{"doc_id": "doc_000195", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Saffron had a method update on 2026-05-10."}, {"sent_id": 3, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}, {"sent_id": 4, "text": "Project Meridian logged a personnel update on 2026-06-16: Nadia Chen took the evaluation owner responsibilities."}, {"sent_id": 5, "text": "Milestone P1 for Project Meridian was revised on 2026-04-03."}, {"sent_id": 6, "text": "The schedule moved from 2026-04-11 to 2026-04-15."}, {"sent_id": 7, "text": "In the 2026-06-22 staffing section, Project Sonata listed Dr. Jonas Nolan as the retrieval owner."}, {"sent_id": 8, "text": "Milestone D1 for Project Nereid was revised on 2026-05-07."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-14 to 2026-05-13."}, {"sent_id": 10, "text": "Quartz-7B was the model selected by Project Saffron on 2026-04-02 for the claim classification track."}], "source": "synthetic_rewritten", "title": "Project Update 195"}
{"doc_id": "doc_000196", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata received 1 GPU from Node Sycamore-13 in the 2026-06-06 allocation log."}, {"sent_id": 3, "text": "A compute allocation for Project Aster was logged on 2026-06-20."}, {"sent_id": 4, "text": "Node Spruce-03 supplied 3 GPUs."}, {"sent_id": 5, "text": "The 2026-06-20 method entry for Project Nereid moved from structured prompting to QLoRA adaptation."}, {"sent_id": 6, "text": "For calibration, the 2026-04-10 note says Project Sonata went with Mira-8B."}, {"sent_id": 7, "text": "The 2026-05-27 compute note links Node Aspen-01 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 8, "text": "Aster-3B was the model selected by Project Sonata on 2026-05-22 for the evidence retrieval track."}, {"sent_id": 9, "text": "Project Nereid had a failed run on 2026-05-16 involving Atlas-3B and Node Hazel-14."}, {"sent_id": 10, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 11, "text": "The 2026-04-02 compute note links Node Willow-05 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 12, "text": "Project Nereid's Milestone V1 deadline change on 2026-06-17 replaced 2026-06-28 with 2026-06-23."}], "source": "synthetic_rewritten", "title": "Lab Digest 196"}
{"doc_id": "doc_000197", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian had a method update on 2026-06-14."}, {"sent_id": 3, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 4, "text": "For error analysis, the 2026-05-21 note says Project Anchor went with Finch-8B."}, {"sent_id": 5, "text": "A compute allocation for Project Saffron was logged on 2026-06-13."}, {"sent_id": 6, "text": "Node Sycamore-13 supplied 4 GPUs."}, {"sent_id": 7, "text": "A compute allocation for Project Meridian was logged on 2026-06-06."}, {"sent_id": 8, "text": "Node Laurel-11 supplied 2 GPUs."}, {"sent_id": 9, "text": "The staffing note for Project Anchor was dated 2026-06-22."}, {"sent_id": 10, "text": "It named Kira Iyer for the evaluation owner duties."}, {"sent_id": 11, "text": "Milestone X1 for Project Sonata was revised on 2026-05-22."}, {"sent_id": 12, "text": "The schedule moved from 2026-05-29 to 2026-05-28."}, {"sent_id": 13, "text": "Project Meridian had a method update on 2026-05-31."}, {"sent_id": 14, "text": "The old approach was reward reranking, and the replacement was data mixing."}, {"sent_id": 15, "text": "Project Saffron had a method update on 2026-06-18."}, {"sent_id": 16, "text": "The old approach was sentence pruning, and the replacement was cross-encoder reranking."}], "source": "synthetic_rewritten", "title": "Project Update 197"}
{"doc_id": "doc_000198", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Anchor had a method update on 2026-06-06."}, {"sent_id": 3, "text": "The old approach was rank fusion, and the replacement was teacher distillation."}, {"sent_id": 4, "text": "For calibration, the 2026-04-29 note says Project Aster went with Lumen-3B."}, {"sent_id": 5, "text": "On 2026-06-05, Project Anchor shifted Milestone Z1: 2026-06-23 became 2026-06-29."}, {"sent_id": 6, "text": "Project Nereid's Milestone D1 deadline change on 2026-06-17 replaced 2026-07-04 with 2026-07-09."}, {"sent_id": 7, "text": "Project Anchor recorded a model choice for evidence retrieval on 2026-05-14."}, {"sent_id": 8, "text": "The chosen model was River-7B."}, {"sent_id": 9, "text": "The 2026-06-20 result for Project Meridian put Atlas-8B on SignalSet-3 at 0.759 latency efficiency score, with Node Hazel-14 handling the run."}, {"sent_id": 10, "text": "Project Meridian had a failed run on 2026-06-11 involving Nimbus-7B and Node Elm-08."}, {"sent_id": 11, "text": "The recorded cause was a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Lab Digest 198"}
{"doc_id": "doc_000199", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata had a method update on 2026-04-15."}, {"sent_id": 3, "text": "The old approach was document chunking, and the replacement was metric smoothing."}, {"sent_id": 4, "text": "On 2026-04-09, Project Nereid shifted Milestone V1: 2026-04-30 became 2026-05-09."}, {"sent_id": 5, "text": "A compute allocation for Project Nereid was logged on 2026-04-10."}, {"sent_id": 6, "text": "Node Birch-04 supplied 1 GPU."}, {"sent_id": 7, "text": "The 2026-05-30 compute note links Node Hazel-14 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 8, "text": "Project Anchor logged a personnel update on 2026-05-18: Leo Park took the evaluation owner responsibilities."}, {"sent_id": 9, "text": "The staffing note for Project Saffron was dated 2026-06-16."}, {"sent_id": 10, "text": "It named Owen Torres for the evaluation owner duties."}, {"sent_id": 11, "text": "Nova-3B was the model selected by Project Nereid on 2026-04-03 for the reranking track."}], "source": "synthetic_rewritten", "title": "Project Update 199"}
{"doc_id": "doc_000200", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Aster recorded a model choice for evidence retrieval on 2026-06-30."}, {"sent_id": 3, "text": "The chosen model was River-8B."}, {"sent_id": 4, "text": "On 2026-05-06, Project Meridian retired reward reranking in favor of data mixing."}, {"sent_id": 5, "text": "Project Sonata had a method update on 2026-05-24."}, {"sent_id": 6, "text": "The old approach was alias expansion, and the replacement was dense retrieval."}, {"sent_id": 7, "text": "For Project Meridian, the run using Kestrel-7B on Node Birch-04 did not complete on 2026-04-17; the issue was a checkpoint-mismatch error."}, {"sent_id": 8, "text": "For Project Nereid, the run using Atlas-8B on Node Hazel-14 did not complete on 2026-05-14; the issue was an unstable-validation-loss error."}, {"sent_id": 9, "text": "Project Aster had a failed run on 2026-04-02 involving Helix-8B and Node Poplar-12."}, {"sent_id": 10, "text": "The recorded cause was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Review Packet 200"}
{"doc_id": "doc_000201", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for claim classification on 2026-06-26."}, {"sent_id": 3, "text": "The chosen model was Kestrel-8B."}, {"sent_id": 4, "text": "The 2026-05-10 method entry for Project Meridian moved from metric smoothing to hard-negative mining."}, {"sent_id": 5, "text": "Project Aster had a method update on 2026-05-23."}, {"sent_id": 6, "text": "The old approach was sentence pruning, and the replacement was cross-encoder reranking."}, {"sent_id": 7, "text": "The staffing note for Project Anchor was dated 2026-06-22."}, {"sent_id": 8, "text": "It named Dr. Arun Kwan for the data steward duties."}, {"sent_id": 9, "text": "Project Nereid recorded a model choice for calibration on 2026-05-27."}, {"sent_id": 10, "text": "The chosen model was Nimbus-3B."}, {"sent_id": 11, "text": "On 2026-05-03, Project Sonata retired evidence pooling in favor of calibrated voting."}, {"sent_id": 12, "text": "On 2026-06-07, Project Meridian retired BM25 retrieval in favor of evidence pooling."}, {"sent_id": 13, "text": "Project Sonata recorded a model choice for error analysis on 2026-06-05."}, {"sent_id": 14, "text": "The chosen model was Marble-8B."}, {"sent_id": 15, "text": "The 2026-06-17 method entry for Project Aster moved from hard-negative mining to LoRA adaptation."}], "source": "synthetic_rewritten", "title": "Run Brief 201"}
{"doc_id": "doc_000202", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Anchor had a failed run on 2026-04-09 involving River-3B and Node Rowan-09."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "Project Sonata logged a personnel update on 2026-06-02: Mara Lane took the retrieval owner responsibilities."}, {"sent_id": 5, "text": "Project Anchor ran River-8B on SignalSet-2 using Node Rowan-09 on 2026-05-16."}, {"sent_id": 6, "text": "The reported latency efficiency score was 0.869."}, {"sent_id": 7, "text": "The staffing note for Project Saffron was dated 2026-04-28."}, {"sent_id": 8, "text": "It named Petra Adler for the data steward duties."}, {"sent_id": 9, "text": "Using Node Sycamore-13, Mira-8B produced 0.553 latency efficiency score on MemoTrace-2 for Project Saffron on 2026-05-09."}, {"sent_id": 10, "text": "The 2026-04-24 compute note links Node Poplar-12 to Project Anchor with an allocation of 4 GPUs."}], "source": "synthetic_rewritten", "title": "Review Packet 202"}
{"doc_id": "doc_000203", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata ran Aster-7B on RiverBench-3 using Node Willow-05 on 2026-05-12."}, {"sent_id": 3, "text": "The reported evidence F1 was 0.694."}, {"sent_id": 4, "text": "For claim classification, the 2026-06-28 note says Project Anchor went with Helix-7B."}, {"sent_id": 5, "text": "Project Meridian's Milestone D1 deadline change on 2026-05-04 replaced 2026-05-26 with 2026-06-05."}, {"sent_id": 6, "text": "Project Sonata had a method update on 2026-04-23."}, {"sent_id": 7, "text": "The old approach was dense retrieval, and the replacement was rank fusion."}, {"sent_id": 8, "text": "River-7B was the model selected by Project Anchor on 2026-05-24 for the evidence retrieval track."}, {"sent_id": 9, "text": "In the 2026-06-02 staffing section, Project Anchor listed Selene Rios as the data steward."}], "source": "synthetic_rewritten", "title": "Run Brief 203"}
{"doc_id": "doc_000204", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor's Milestone B1 deadline change on 2026-04-24 replaced 2026-05-20 with 2026-05-26."}, {"sent_id": 3, "text": "Project Aster had a failed run on 2026-04-09 involving Orchid-8B and Node Cedar-02."}, {"sent_id": 4, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 5, "text": "On 2026-06-12, Project Nereid shifted Milestone B2: 2026-07-01 became 2026-06-28."}, {"sent_id": 6, "text": "A compute allocation for Project Saffron was logged on 2026-04-25."}, {"sent_id": 7, "text": "Node Fir-10 supplied 2 GPUs."}, {"sent_id": 8, "text": "Project Sonata received 1 GPU from Node Maple-01 in the 2026-05-02 allocation log."}, {"sent_id": 9, "text": "The 2026-05-23 method entry for Project Meridian moved from chain verification to structured prompting."}, {"sent_id": 10, "text": "Project Anchor's Milestone H1 deadline change on 2026-05-04 replaced 2026-05-28 with 2026-06-05."}, {"sent_id": 11, "text": "Project Nereid recorded a model choice for calibration on 2026-06-14."}, {"sent_id": 12, "text": "The chosen model was Nimbus-3B."}], "source": "synthetic_rewritten", "title": "Review Packet 204"}
{"doc_id": "doc_000205", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-06-27 method entry for Project Aster moved from reward reranking to data mixing."}, {"sent_id": 3, "text": "For Project Anchor, the run using Lumen-8B on Node Spruce-03 did not complete on 2026-04-04; the issue was a missing-index error."}, {"sent_id": 4, "text": "The 2026-04-02 failure note says Atlas-3B on Node Hazel-14 failed for Project Meridian because of a checkpoint-mismatch error."}, {"sent_id": 5, "text": "Project Anchor logged a personnel update on 2026-04-28: Iris Lane took the evaluation owner responsibilities."}, {"sent_id": 6, "text": "For Project Saffron, the run using Vela-3B on Node Fir-10 did not complete on 2026-04-30; the issue was a missing-index error."}, {"sent_id": 7, "text": "On 2026-04-09, Project Meridian shifted Milestone P1: 2026-04-17 became 2026-04-25."}, {"sent_id": 8, "text": "Project Meridian recorded a model choice for claim classification on 2026-04-08."}, {"sent_id": 9, "text": "The chosen model was Kestrel-7B."}, {"sent_id": 10, "text": "The 2026-05-09 result for Project Anchor put Helix-8B on CedarQA-3 at 0.527 latency efficiency score, with Node Poplar-12 handling the run."}, {"sent_id": 11, "text": "The review note for Amber Ranking Study 4 was dated 2026-04-15."}, {"sent_id": 12, "text": "It described the method as LoRA adaptation and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Run Brief 205"}
{"doc_id": "doc_000206", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "A compute allocation for Project Anchor was logged on 2026-04-18."}, {"sent_id": 3, "text": "Node Rowan-09 supplied 4 GPUs."}, {"sent_id": 4, "text": "For Project Sonata, the run using Mira-3B on Node Sycamore-13 did not complete on 2026-06-18; the issue was an out-of-memory error."}, {"sent_id": 5, "text": "River-3B was the model selected by Project Aster on 2026-04-29 for the evidence retrieval track."}, {"sent_id": 6, "text": "Project Aster received 1 GPU from Node Cedar-02 in the 2026-06-08 allocation log."}, {"sent_id": 7, "text": "The 2026-04-01 method entry for Project Sonata moved from structured prompting to QLoRA adaptation."}, {"sent_id": 8, "text": "On 2026-05-18, Project Saffron shifted Milestone L1: 2026-06-13 became 2026-06-23."}], "source": "synthetic_rewritten", "title": "Review Packet 206"}
{"doc_id": "doc_000207", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Orchid-3B was the model selected by Project Aster on 2026-04-08 for the reranking track."}, {"sent_id": 3, "text": "The 2026-05-11 result for Project Sonata put Marble-3B on NereidNotes at 0.556 evidence F1, with Node Maple-01 handling the run."}, {"sent_id": 4, "text": "Finch-8B was the model selected by Project Aster on 2026-04-08 for the error analysis track."}, {"sent_id": 5, "text": "Project Anchor's Milestone B1 deadline change on 2026-04-23 replaced 2026-05-09 with 2026-05-13."}, {"sent_id": 6, "text": "In the 2026-05-05 staffing section, Project Nereid listed Nora Bauer as the lead."}, {"sent_id": 7, "text": "On 2026-06-27, Project Nereid retired evidence pooling in favor of calibrated voting."}, {"sent_id": 8, "text": "Project Meridian had a failed run on 2026-06-26 involving Nimbus-7B and Node Elm-08."}, {"sent_id": 9, "text": "The recorded cause was a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 207"}
{"doc_id": "doc_000208", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-04-07 staffing section, Project Nereid listed Vera Torres as the retrieval owner."}, {"sent_id": 3, "text": "The 2026-04-18 compute note links Node Aspen-01 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 4, "text": "The 2026-05-05 result for Project Saffron put Vela-7B on LumenFacts-3 at 0.625 accuracy, with Node Fir-10 handling the run."}, {"sent_id": 5, "text": "The 2026-06-05 failure note says Orchid-7B on Node Cedar-02 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 6, "text": "On 2026-05-15, Project Nereid shifted Milestone P1: 2026-05-28 became 2026-06-02."}, {"sent_id": 7, "text": "The 2026-06-15 result for Project Meridian put Nimbus-3B on MemoTrace at 0.693 accuracy, with Node Elm-08 handling the run."}, {"sent_id": 8, "text": "Project Sonata received 1 GPU from Node Sycamore-13 in the 2026-04-27 allocation log."}, {"sent_id": 9, "text": "Project Saffron recorded a model choice for calibration on 2026-05-24."}, {"sent_id": 10, "text": "The chosen model was Mira-3B."}], "source": "synthetic_rewritten", "title": "Review Packet 208"}
{"doc_id": "doc_000209", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Saffron logged a personnel update on 2026-04-21: Dr. Theo Grant took the data steward responsibilities."}, {"sent_id": 3, "text": "For Project Nereid, the run using Kestrel-8B on Node Birch-04 did not complete on 2026-04-30; the issue was an out-of-memory error."}, {"sent_id": 4, "text": "Milestone F1 for Project Sonata was revised on 2026-06-11."}, {"sent_id": 5, "text": "The schedule moved from 2026-06-24 to 2026-07-03."}, {"sent_id": 6, "text": "For Project Saffron, the run using Quartz-8B on Node Pine-07 did not complete on 2026-04-11; the issue was a missing-index error."}, {"sent_id": 7, "text": "Project Aster recorded a model choice for calibration on 2026-05-15."}, {"sent_id": 8, "text": "The chosen model was Lumen-7B."}, {"sent_id": 9, "text": "In the 2026-06-07 staffing section, Project Sonata listed Sofia Nadir as the retrieval owner."}, {"sent_id": 10, "text": "On 2026-05-13, Project Nereid retired LoRA adaptation in favor of alias expansion."}], "source": "synthetic_rewritten", "title": "Run Brief 209"}
{"doc_id": "doc_000210", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-16 compute note links Node Rowan-09 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 3, "text": "Project Anchor recorded a model choice for evidence retrieval on 2026-06-24."}, {"sent_id": 4, "text": "The chosen model was River-3B."}, {"sent_id": 5, "text": "The 2026-05-16 compute note links Node Birch-04 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 6, "text": "Orchid-8B was the model selected by Project Aster on 2026-06-24 for the reranking track."}, {"sent_id": 7, "text": "Milestone F1 for Project Sonata was revised on 2026-04-29."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-12 to 2026-05-21."}, {"sent_id": 9, "text": "In the 2026-05-12 staffing section, Project Nereid listed Darian Grant as the lead."}], "source": "synthetic_rewritten", "title": "Lab Digest 210"}
{"doc_id": "doc_000211", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian recorded a model choice for claim classification on 2026-06-04."}, {"sent_id": 3, "text": "The chosen model was Kestrel-7B."}, {"sent_id": 4, "text": "The 2026-04-07 result for Project Anchor put Helix-7B on CedarQA-3 at 0.557 latency efficiency score, with Node Poplar-12 handling the run."}, {"sent_id": 5, "text": "The 2026-04-26 method entry for Project Sonata moved from structured prompting to QLoRA adaptation."}, {"sent_id": 6, "text": "Project Nereid ran Kestrel-3B on OrionBench-3 using Node Birch-04 on 2026-06-16."}, {"sent_id": 7, "text": "The reported evidence F1 was 0.792."}, {"sent_id": 8, "text": "A compute allocation for Project Anchor was logged on 2026-06-27."}, {"sent_id": 9, "text": "Node Rowan-09 supplied 2 GPUs."}, {"sent_id": 10, "text": "The 2026-06-01 result for Project Anchor put Lumen-7B on MemoTrace-3 at 0.893 latency efficiency score, with Node Spruce-03 handling the run."}, {"sent_id": 11, "text": "Project Anchor logged a personnel update on 2026-04-14: Kira Frost took the data steward responsibilities."}], "source": "synthetic_rewritten", "title": "Project Update 211"}
{"doc_id": "doc_000212", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Milestone N1 for Project Aster was revised on 2026-04-02."}, {"sent_id": 3, "text": "The schedule moved from 2026-04-09 to 2026-04-08."}, {"sent_id": 4, "text": "Project Nereid logged a personnel update on 2026-04-13: Vera Torres took the retrieval owner responsibilities."}, {"sent_id": 5, "text": "Project Sonata had a method update on 2026-06-04."}, {"sent_id": 6, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 7, "text": "The staffing note for Project Sonata was dated 2026-04-13."}, {"sent_id": 8, "text": "It named Sofia Nadir for the retrieval owner duties."}, {"sent_id": 9, "text": "The 2026-04-16 failure note says Marble-8B on Node Maple-01 failed for Project Saffron because of a missing-index error."}, {"sent_id": 10, "text": "Project Anchor received 4 GPUs from Node Juniper-06 in the 2026-05-09 allocation log."}, {"sent_id": 11, "text": "Project Nereid ran Nova-8B on LabQA using Node Aspen-01 on 2026-05-11."}, {"sent_id": 12, "text": "The reported macro F1 was 0.660."}, {"sent_id": 13, "text": "On 2026-04-11, Project Saffron retired contrastive tuning in favor of threshold search."}, {"sent_id": 14, "text": "Project Meridian received 4 GPUs from Node Hazel-14 in the 2026-06-05 allocation log."}], "source": "synthetic_rewritten", "title": "Lab Digest 212"}
{"doc_id": "doc_000213", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian had a failed run on 2026-05-09 involving Nimbus-7B and Node Elm-08."}, {"sent_id": 3, "text": "The recorded cause was a missing-index error."}, {"sent_id": 4, "text": "Using Node Hazel-14, Atlas-3B produced 0.819 accuracy on SignalSet-3 for Project Meridian on 2026-05-26."}, {"sent_id": 5, "text": "For evidence retrieval, the 2026-05-14 note says Project Sonata went with Aster-8B."}, {"sent_id": 6, "text": "The staffing note for Project Aster was dated 2026-05-12."}, {"sent_id": 7, "text": "It named Dr. Noah Vale for the lead duties."}, {"sent_id": 8, "text": "Project Nereid ran Kestrel-7B on OrionBench-3 using Node Birch-04 on 2026-06-27."}, {"sent_id": 9, "text": "The reported macro F1 was 0.672."}, {"sent_id": 10, "text": "On 2026-06-26, Project Nereid shifted Milestone J1: 2026-07-21 became 2026-07-26."}, {"sent_id": 11, "text": "The staffing note for Project Anchor was dated 2026-05-05."}, {"sent_id": 12, "text": "It named Kira Iyer for the evaluation owner duties."}, {"sent_id": 13, "text": "The 2026-04-03 failure note says Vela-7B on Node Fir-10 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 14, "text": "The 2026-04-18 method entry for Project Saffron moved from reward reranking to data mixing."}], "source": "synthetic_rewritten", "title": "Project Update 213"}
{"doc_id": "doc_000214", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian recorded a model choice for claim classification on 2026-05-28."}, {"sent_id": 3, "text": "The chosen model was Kestrel-8B."}, {"sent_id": 4, "text": "Milestone H1 for Project Aster was revised on 2026-06-23."}, {"sent_id": 5, "text": "The schedule moved from 2026-07-02 to 2026-07-11."}, {"sent_id": 6, "text": "Aster-7B was the model selected by Project Sonata on 2026-05-31 for the evidence retrieval track."}, {"sent_id": 7, "text": "The staffing note for Project Meridian was dated 2026-04-13."}, {"sent_id": 8, "text": "It named Camila Quinn for the evaluation owner duties."}, {"sent_id": 9, "text": "Project Sonata logged a personnel update on 2026-04-14: Sofia Frost took the lead responsibilities."}, {"sent_id": 10, "text": "Quartz-8B was the model selected by Project Sonata on 2026-05-15 for the claim classification track."}, {"sent_id": 11, "text": "Milestone T1 for Project Anchor was revised on 2026-04-17."}, {"sent_id": 12, "text": "The schedule moved from 2026-04-27 to 2026-05-03."}], "source": "synthetic_rewritten", "title": "Lab Digest 214"}
{"doc_id": "doc_000215", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The staffing note for Project Saffron was dated 2026-04-14."}, {"sent_id": 3, "text": "It named Owen Marin for the data steward duties."}, {"sent_id": 4, "text": "River-8B was the model selected by Project Anchor on 2026-05-17 for the evidence retrieval track."}, {"sent_id": 5, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-04-22."}, {"sent_id": 6, "text": "The chosen model was Atlas-3B."}, {"sent_id": 7, "text": "Project Sonata logged a personnel update on 2026-04-06: Dr. Jonas Nolan took the retrieval owner responsibilities."}, {"sent_id": 8, "text": "For calibration, the 2026-06-05 note says Project Anchor went with Lumen-3B."}, {"sent_id": 9, "text": "Project Aster logged a personnel update on 2026-06-29: Dr. Noah Vale took the lead responsibilities."}, {"sent_id": 10, "text": "Project Saffron had a method update on 2026-04-26."}, {"sent_id": 11, "text": "The old approach was query rewriting, and the replacement was sentence pruning."}], "source": "synthetic_rewritten", "title": "Project Update 215"}
{"doc_id": "doc_000216", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata received 1 GPU from Node Willow-05 in the 2026-06-17 allocation log."}, {"sent_id": 3, "text": "For Project Nereid, the run using Kestrel-3B on Node Birch-04 did not complete on 2026-06-04; the issue was an unstable-validation-loss error."}, {"sent_id": 4, "text": "The 2026-05-21 compute note links Node Birch-04 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 5, "text": "Project Aster recorded a model choice for error analysis on 2026-04-09."}, {"sent_id": 6, "text": "The chosen model was Finch-3B."}, {"sent_id": 7, "text": "The 2026-05-26 result for Project Meridian put Cedar-7B on VestaLogs-2 at 0.741 latency efficiency score, with Node Laurel-11 handling the run."}, {"sent_id": 8, "text": "The 2026-05-21 failure note says Mira-7B on Node Sycamore-13 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 9, "text": "For reranking, the 2026-06-25 note says Project Nereid went with Nova-8B."}], "source": "synthetic_rewritten", "title": "Lab Digest 216"}
{"doc_id": "doc_000217", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Helix-8B was the model selected by Project Anchor on 2026-04-28 for the claim classification track."}, {"sent_id": 3, "text": "Project Aster had a failed run on 2026-05-21 involving River-8B and Node Rowan-09."}, {"sent_id": 4, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 5, "text": "On 2026-05-04, Project Aster shifted Milestone B1: 2026-05-15 became 2026-05-10."}, {"sent_id": 6, "text": "A compute allocation for Project Aster was logged on 2026-05-30."}, {"sent_id": 7, "text": "Node Rowan-09 supplied 1 GPU."}, {"sent_id": 8, "text": "On 2026-05-27, Project Nereid shifted Milestone J1: 2026-06-11 became 2026-06-07."}, {"sent_id": 9, "text": "The 2026-05-23 result for Project Nereid put Nimbus-3B on TraceEval-2 at 0.678 evidence F1, with Node Elm-08 handling the run."}, {"sent_id": 10, "text": "The 2026-05-23 compute note links Node Juniper-06 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 11, "text": "The 2026-04-03 compute note links Node Willow-05 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 12, "text": "The review note for Stable Chains Study 2 was dated 2026-04-15."}, {"sent_id": 13, "text": "It described the method as chain verification and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Project Update 217"}
{"doc_id": "doc_000218", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Aster logged a personnel update on 2026-04-20: Talia Reed took the retrieval owner responsibilities."}, {"sent_id": 3, "text": "Project Sonata logged a personnel update on 2026-04-12: Mara Quinn took the lead responsibilities."}, {"sent_id": 4, "text": "The 2026-05-25 result for Project Anchor put Orchid-3B on LumenFacts at 0.785 latency efficiency score, with Node Cedar-02 handling the run."}, {"sent_id": 5, "text": "The staffing note for Project Sonata was dated 2026-05-12."}, {"sent_id": 6, "text": "It named Rohan Singh for the retrieval owner duties."}, {"sent_id": 7, "text": "Project Saffron received 4 GPUs from Node Pine-07 in the 2026-05-23 allocation log."}, {"sent_id": 8, "text": "The 2026-05-21 failure note says Helix-7B on Node Poplar-12 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 9, "text": "For Project Sonata, the run using Aster-8B on Node Willow-05 did not complete on 2026-06-20; the issue was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Lab Digest 218"}
{"doc_id": "doc_000219", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For calibration, the 2026-06-28 note says Project Anchor went with Lumen-8B."}, {"sent_id": 3, "text": "Using Node Pine-07, Quartz-3B produced 0.562 macro F1 on OrionBench for Project Sonata on 2026-05-02."}, {"sent_id": 4, "text": "In the 2026-04-26 staffing section, Project Nereid listed Dr. Julian Stone as the retrieval owner."}, {"sent_id": 5, "text": "Project Nereid recorded a model choice for claim classification on 2026-04-08."}, {"sent_id": 6, "text": "The chosen model was Kestrel-3B."}, {"sent_id": 7, "text": "The 2026-04-23 failure note says Kestrel-7B on Node Birch-04 failed for Project Meridian because of a checkpoint-mismatch error."}, {"sent_id": 8, "text": "On 2026-04-19, Project Sonata retired metric smoothing in favor of hard-negative mining."}, {"sent_id": 9, "text": "Using Node Cedar-02, Orchid-8B produced 0.815 accuracy on LumenFacts for Project Anchor on 2026-06-01."}], "source": "synthetic_rewritten", "title": "Project Update 219"}
{"doc_id": "doc_000220", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Using Node Willow-05, Aster-8B produced 0.724 evidence F1 on RiverBench-3 for Project Sonata on 2026-06-14."}, {"sent_id": 3, "text": "Using Node Sycamore-13, Mira-8B produced 0.643 latency efficiency score on MemoTrace-2 for Project Saffron on 2026-04-21."}, {"sent_id": 4, "text": "Project Nereid's Milestone V1 deadline change on 2026-04-16 replaced 2026-04-27 with 2026-04-22."}, {"sent_id": 5, "text": "Milestone T1 for Project Anchor was revised on 2026-04-10."}, {"sent_id": 6, "text": "The schedule moved from 2026-04-30 to 2026-05-08."}, {"sent_id": 7, "text": "Project Anchor had a failed run on 2026-05-07 involving Finch-7B and Node Juniper-06."}, {"sent_id": 8, "text": "The recorded cause was a missing-index error."}, {"sent_id": 9, "text": "Using Node Rowan-09, River-8B produced 0.869 accuracy on SignalSet-2 for Project Anchor on 2026-04-07."}], "source": "synthetic_rewritten", "title": "Run Brief 220"}
{"doc_id": "doc_000221", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Milestone H1 for Project Aster was revised on 2026-05-01."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-10 to 2026-05-15."}, {"sent_id": 4, "text": "The 2026-05-15 failure note says Nimbus-8B on Node Elm-08 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 5, "text": "On 2026-05-30, Project Aster retired dense retrieval in favor of rank fusion."}, {"sent_id": 6, "text": "A compute allocation for Project Saffron was logged on 2026-04-10."}, {"sent_id": 7, "text": "Node Pine-07 supplied 2 GPUs."}, {"sent_id": 8, "text": "For reranking, the 2026-04-09 note says Project Sonata went with Vela-7B."}, {"sent_id": 9, "text": "Project Nereid had a failed run on 2026-06-20 involving Nimbus-3B and Node Elm-08."}, {"sent_id": 10, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 11, "text": "Project Aster ran Orchid-3B on LabQA-3 using Node Cedar-02 on 2026-05-05."}, {"sent_id": 12, "text": "The reported evidence F1 was 0.560."}, {"sent_id": 13, "text": "Project Sonata logged a personnel update on 2026-05-12: Samir Ames took the lead responsibilities."}, {"sent_id": 14, "text": "Project Aster recorded a model choice for reranking on 2026-04-12."}, {"sent_id": 15, "text": "The chosen model was Orchid-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 221"}
{"doc_id": "doc_000222", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-02 compute note links Node Rowan-09 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 3, "text": "For error analysis, the 2026-05-10 note says Project Saffron went with Marble-3B."}, {"sent_id": 4, "text": "For reranking, the 2026-05-21 note says Project Aster went with Orchid-8B."}, {"sent_id": 5, "text": "Project Anchor ran River-7B on SignalSet-2 using Node Rowan-09 on 2026-06-02."}, {"sent_id": 6, "text": "The reported latency efficiency score was 0.719."}, {"sent_id": 7, "text": "On 2026-04-10, the Amber Ranking Study 1 review reported LoRA adaptation and said it did not use a reward model."}, {"sent_id": 8, "text": "Project Saffron's Milestone X1 deadline change on 2026-04-02 replaced 2026-04-24 with 2026-04-30."}, {"sent_id": 9, "text": "Orchid-7B was the model selected by Project Aster on 2026-05-26 for the reranking track."}, {"sent_id": 10, "text": "Milestone D1 for Project Nereid was revised on 2026-06-10."}, {"sent_id": 11, "text": "The schedule moved from 2026-06-17 to 2026-06-16."}, {"sent_id": 12, "text": "Project Sonata had a method update on 2026-06-14."}, {"sent_id": 13, "text": "The old approach was QLoRA adaptation, and the replacement was hybrid retrieval."}], "source": "synthetic_rewritten", "title": "Run Brief 222"}
{"doc_id": "doc_000223", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian ran Atlas-3B on SignalSet-3 using Node Hazel-14 on 2026-05-05."}, {"sent_id": 3, "text": "The reported latency efficiency score was 0.729."}, {"sent_id": 4, "text": "Aster-3B was the model selected by Project Saffron on 2026-04-02 for the evidence retrieval track."}, {"sent_id": 5, "text": "For Project Saffron, the run using Mira-8B on Node Sycamore-13 did not complete on 2026-04-09; the issue was a missing-index error."}, {"sent_id": 6, "text": "Using Node Pine-07, Quartz-7B produced 0.592 evidence F1 on OrionBench for Project Sonata on 2026-04-13."}, {"sent_id": 7, "text": "On 2026-05-08, the Pale Compass Study 3 review reported temporal filtering and said it did not use a reward model."}, {"sent_id": 8, "text": "Atlas-7B was the model selected by Project Nereid on 2026-05-08 for the evidence retrieval track."}, {"sent_id": 9, "text": "For claim classification, the 2026-04-24 note says Project Nereid went with Kestrel-7B."}, {"sent_id": 10, "text": "The 2026-06-27 method entry for Project Saffron moved from alias expansion to dense retrieval."}], "source": "synthetic_rewritten", "title": "Review Packet 223"}
{"doc_id": "doc_000224", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor recorded a model choice for calibration on 2026-05-24."}, {"sent_id": 3, "text": "The chosen model was Lumen-3B."}, {"sent_id": 4, "text": "The staffing note for Project Nereid was dated 2026-05-25."}, {"sent_id": 5, "text": "It named Dr. Julian Gray for the lead duties."}, {"sent_id": 6, "text": "Using Node Pine-07, Quartz-3B produced 0.532 evidence F1 on OrionBench for Project Sonata on 2026-06-29."}, {"sent_id": 7, "text": "Quartz-8B was the model selected by Project Saffron on 2026-04-29 for the claim classification track."}, {"sent_id": 8, "text": "Nimbus-3B was the model selected by Project Nereid on 2026-05-17 for the calibration track."}, {"sent_id": 9, "text": "The 2026-05-20 method entry for Project Sonata moved from QLoRA adaptation to hybrid retrieval."}, {"sent_id": 10, "text": "Quartz-7B was the model selected by Project Saffron on 2026-06-24 for the claim classification track."}], "source": "synthetic_rewritten", "title": "Run Brief 224"}
{"doc_id": "doc_000225", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For reranking, the 2026-05-17 note says Project Nereid went with Nova-8B."}, {"sent_id": 3, "text": "The 2026-06-15 compute note links Node Willow-05 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 4, "text": "In the 2026-06-23 staffing section, Project Aster listed Anika Costa as the retrieval owner."}, {"sent_id": 5, "text": "On 2026-06-10, Project Aster retired calibrated voting in favor of confidence calibration."}, {"sent_id": 6, "text": "A compute allocation for Project Anchor was logged on 2026-05-23."}, {"sent_id": 7, "text": "Node Spruce-03 supplied 2 GPUs."}, {"sent_id": 8, "text": "The 2026-06-04 method entry for Project Nereid moved from confidence calibration to chain verification."}], "source": "synthetic_rewritten", "title": "Review Packet 225"}
{"doc_id": "doc_000226", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-26, Project Saffron retired temporal filtering in favor of alias expansion."}, {"sent_id": 3, "text": "The 2026-06-21 method entry for Project Meridian moved from hard-negative mining to LoRA adaptation."}, {"sent_id": 4, "text": "The 2026-06-01 result for Project Nereid put Nova-8B on LabQA at 0.750 evidence F1, with Node Aspen-01 handling the run."}, {"sent_id": 5, "text": "For reranking, the 2026-04-16 note says Project Anchor went with Orchid-8B."}, {"sent_id": 6, "text": "On 2026-05-01, Project Nereid shifted Milestone D1: 2026-05-08 became 2026-05-07."}, {"sent_id": 7, "text": "The 2026-06-11 compute note links Node Spruce-03 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 8, "text": "Milestone D2 for Project Sonata was revised on 2026-04-16."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-01 to 2026-04-27."}, {"sent_id": 10, "text": "The 2026-04-13 compute note links Node Aspen-01 to Project Nereid with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Run Brief 226"}
{"doc_id": "doc_000227", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-04-28 staffing section, Project Aster listed Dr. Noah Vale as the lead."}, {"sent_id": 3, "text": "The 2026-06-11 method entry for Project Nereid moved from hybrid retrieval to document chunking."}, {"sent_id": 4, "text": "On 2026-04-12, Project Anchor retired query rewriting in favor of sentence pruning."}, {"sent_id": 5, "text": "The staffing note for Project Anchor was dated 2026-04-05."}, {"sent_id": 6, "text": "It named Leo Hale for the data steward duties."}, {"sent_id": 7, "text": "On 2026-05-27, Project Aster shifted Milestone Z1: 2026-06-19 became 2026-06-17."}, {"sent_id": 8, "text": "For Project Anchor, the run using Finch-3B on Node Juniper-06 did not complete on 2026-05-14; the issue was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "The 2026-04-08 method entry for Project Sonata moved from query rewriting to sentence pruning."}, {"sent_id": 10, "text": "Project Aster ran Helix-3B on OrionBench-2 using Node Poplar-12 on 2026-04-28."}, {"sent_id": 11, "text": "The reported evidence F1 was 0.842."}, {"sent_id": 12, "text": "Project Aster had a method update on 2026-06-03."}, {"sent_id": 13, "text": "The old approach was rank fusion, and the replacement was teacher distillation."}], "source": "synthetic_rewritten", "title": "Review Packet 227"}
{"doc_id": "doc_000228", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For reranking, the 2026-06-28 note says Project Nereid went with Nova-3B."}, {"sent_id": 3, "text": "Project Meridian logged a personnel update on 2026-06-02: Mira Nolan took the evaluation owner responsibilities."}, {"sent_id": 4, "text": "On 2026-05-09, Project Nereid retired reward reranking in favor of data mixing."}, {"sent_id": 5, "text": "River-7B was the model selected by Project Anchor on 2026-05-08 for the evidence retrieval track."}, {"sent_id": 6, "text": "The 2026-06-04 failure note says Quartz-8B on Node Pine-07 failed for Project Saffron because of a missing-index error."}, {"sent_id": 7, "text": "The 2026-06-05 failure note says River-8B on Node Rowan-09 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 8, "text": "Project Aster logged a personnel update on 2026-06-01: Felix Brooks took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Run Brief 228"}
{"doc_id": "doc_000229", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For Project Saffron, the run using Marble-3B on Node Maple-01 did not complete on 2026-04-09; the issue was a checkpoint-mismatch error."}, {"sent_id": 3, "text": "The 2026-04-25 result for Project Anchor put Helix-7B on CedarQA-3 at 0.857 latency efficiency score, with Node Poplar-12 handling the run."}, {"sent_id": 4, "text": "Project Sonata's Milestone D2 deadline change on 2026-04-10 replaced 2026-04-25 with 2026-04-21."}, {"sent_id": 5, "text": "For reranking, the 2026-06-25 note says Project Anchor went with Orchid-8B."}, {"sent_id": 6, "text": "Project Anchor logged a personnel update on 2026-06-23: Leo Park took the evaluation owner responsibilities."}, {"sent_id": 7, "text": "Using Node Elm-08, Nimbus-7B produced 0.843 latency efficiency score on MemoTrace for Project Meridian on 2026-05-03."}], "source": "synthetic_rewritten", "title": "Review Packet 229"}
{"doc_id": "doc_000230", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-06-08 staffing section, Project Meridian listed Mira Nolan as the evaluation owner."}, {"sent_id": 3, "text": "The staffing note for Project Sonata was dated 2026-06-15."}, {"sent_id": 4, "text": "It named Samir Kwan for the retrieval owner duties."}, {"sent_id": 5, "text": "Project Aster's Milestone Z1 deadline change on 2026-04-10 replaced 2026-04-23 with 2026-05-02."}, {"sent_id": 6, "text": "Project Sonata ran Vela-8B on LabQA-2 using Node Fir-10 on 2026-04-11."}, {"sent_id": 7, "text": "The reported macro F1 was 0.550."}, {"sent_id": 8, "text": "Project Nereid had a failed run on 2026-04-16 involving Cedar-3B and Node Laurel-11."}, {"sent_id": 9, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 10, "text": "Project Nereid received 1 GPU from Node Birch-04 in the 2026-04-22 allocation log."}, {"sent_id": 11, "text": "For Project Anchor, the run using Lumen-8B on Node Spruce-03 did not complete on 2026-05-28; the issue was a missing-index error."}, {"sent_id": 12, "text": "For evidence retrieval, the 2026-04-22 note says Project Meridian went with Atlas-7B."}, {"sent_id": 13, "text": "Project Anchor had a method update on 2026-05-03."}, {"sent_id": 14, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}], "source": "synthetic_rewritten", "title": "Project Update 230"}
{"doc_id": "doc_000231", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-04-23 failure note says Finch-3B on Node Juniper-06 failed for Project Anchor because of a missing-index error."}, {"sent_id": 3, "text": "The staffing note for Project Sonata was dated 2026-06-08."}, {"sent_id": 4, "text": "It named Dr. Jonas Rios for the lead duties."}, {"sent_id": 5, "text": "On 2026-06-12, Project Anchor shifted Milestone H1: 2026-06-26 became 2026-07-02."}, {"sent_id": 6, "text": "The staffing note for Project Aster was dated 2026-05-26."}, {"sent_id": 7, "text": "It named Anika Sato for the lead duties."}, {"sent_id": 8, "text": "Milestone J1 for Project Nereid was revised on 2026-05-11."}, {"sent_id": 9, "text": "The schedule moved from 2026-06-05 to 2026-06-14."}, {"sent_id": 10, "text": "In the 2026-06-08 staffing section, Project Saffron listed Owen Torres as the evaluation owner."}, {"sent_id": 11, "text": "Project Anchor's Milestone N1 deadline change on 2026-06-29 replaced 2026-07-21 with 2026-07-27."}], "source": "synthetic_rewritten", "title": "Lab Digest 231"}
{"doc_id": "doc_000232", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Sonata was logged on 2026-04-04."}, {"sent_id": 3, "text": "Node Fir-10 supplied 3 GPUs."}, {"sent_id": 4, "text": "Stable Chains Study 4 appeared in a 2026-04-17 review entry using query rewriting; the note said it did not use a reward model."}, {"sent_id": 5, "text": "Project Saffron had a method update on 2026-05-31."}, {"sent_id": 6, "text": "The old approach was late interaction, and the replacement was BM25 retrieval."}, {"sent_id": 7, "text": "The 2026-04-25 result for Project Aster put Lumen-3B on TraceEval at 0.818 macro F1, with Node Spruce-03 handling the run."}, {"sent_id": 8, "text": "The staffing note for Project Anchor was dated 2026-06-29."}, {"sent_id": 9, "text": "It named Selene Rios for the data steward duties."}, {"sent_id": 10, "text": "Project Aster ran River-8B on RiverBench using Node Rowan-09 on 2026-06-08."}, {"sent_id": 11, "text": "The reported evidence F1 was 0.884."}, {"sent_id": 12, "text": "Project Meridian received 2 GPUs from Node Aspen-01 in the 2026-05-25 allocation log."}], "source": "synthetic_rewritten", "title": "Project Update 232"}
{"doc_id": "doc_000233", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-04-21 staffing section, Project Anchor listed Iris Stone as the data steward."}, {"sent_id": 3, "text": "Project Anchor's Milestone H1 deadline change on 2026-05-13 replaced 2026-06-06 with 2026-06-14."}, {"sent_id": 4, "text": "The staffing note for Project Saffron was dated 2026-05-05."}, {"sent_id": 5, "text": "It named Petra Gray for the evaluation owner duties."}, {"sent_id": 6, "text": "For claim classification, the 2026-04-26 note says Project Meridian went with Kestrel-7B."}, {"sent_id": 7, "text": "For Project Anchor, the run using River-7B on Node Rowan-09 did not complete on 2026-04-23; the issue was a checkpoint-mismatch error."}, {"sent_id": 8, "text": "Project Anchor had a method update on 2026-04-12."}, {"sent_id": 9, "text": "The old approach was cross-encoder reranking, and the replacement was late interaction."}, {"sent_id": 10, "text": "Project Aster received 1 GPU from Node Rowan-09 in the 2026-06-26 allocation log."}], "source": "synthetic_rewritten", "title": "Lab Digest 233"}
{"doc_id": "doc_000234", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-04-14 staffing section, Project Saffron listed Dr. Theo Lind as the evaluation owner."}, {"sent_id": 3, "text": "For claim classification, the 2026-04-08 note says Project Aster went with Helix-8B."}, {"sent_id": 4, "text": "Project Nereid received 3 GPUs from Node Elm-08 in the 2026-05-09 allocation log."}, {"sent_id": 5, "text": "For evidence retrieval, the 2026-05-06 note says Project Aster went with River-8B."}, {"sent_id": 6, "text": "Project Meridian received 2 GPUs from Node Aspen-01 in the 2026-06-06 allocation log."}, {"sent_id": 7, "text": "Milestone V1 for Project Meridian was revised on 2026-05-15."}, {"sent_id": 8, "text": "The schedule moved from 2026-06-10 to 2026-06-16."}, {"sent_id": 9, "text": "For Project Sonata, the run using Mira-3B on Node Sycamore-13 did not complete on 2026-06-12; the issue was an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Project Update 234"}
{"doc_id": "doc_000235", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-04-27 staffing section, Project Sonata listed Mara Quinn as the lead."}, {"sent_id": 3, "text": "For Project Saffron, the run using Quartz-7B on Node Pine-07 did not complete on 2026-06-05; the issue was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "Project Nereid logged a personnel update on 2026-05-11: Nora Bauer took the lead responsibilities."}, {"sent_id": 5, "text": "In the 2026-06-30 staffing section, Project Anchor listed Leo Hale as the data steward."}, {"sent_id": 6, "text": "Project Nereid had a failed run on 2026-04-23 involving Kestrel-3B and Node Birch-04."}, {"sent_id": 7, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 8, "text": "Project Nereid received 1 GPU from Node Elm-08 in the 2026-05-16 allocation log."}, {"sent_id": 9, "text": "Helix-8B was the model selected by Project Anchor on 2026-06-03 for the claim classification track."}, {"sent_id": 10, "text": "Nova-7B was the model selected by Project Meridian on 2026-05-07 for the reranking track."}, {"sent_id": 11, "text": "The 2026-05-15 failure note says Atlas-8B on Node Hazel-14 failed for Project Meridian because of a missing-index error."}], "source": "synthetic_rewritten", "title": "Lab Digest 235"}
{"doc_id": "doc_000236", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-19, Project Meridian retired contrastive tuning in favor of threshold search."}, {"sent_id": 3, "text": "Project Sonata's Milestone D2 deadline change on 2026-06-15 replaced 2026-07-10 with 2026-07-15."}, {"sent_id": 4, "text": "The 2026-04-15 method entry for Project Aster moved from late interaction to BM25 retrieval."}, {"sent_id": 5, "text": "On 2026-05-27, Project Sonata shifted Milestone R1: 2026-06-15 became 2026-06-12."}, {"sent_id": 6, "text": "The 2026-05-07 failure note says Finch-3B on Node Juniper-06 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 7, "text": "Cedar-3B was the model selected by Project Nereid on 2026-05-22 for the error analysis track."}, {"sent_id": 8, "text": "The staffing note for Project Sonata was dated 2026-04-28."}, {"sent_id": 9, "text": "It named Samir Ames for the lead duties."}], "source": "synthetic_rewritten", "title": "Project Update 236"}
{"doc_id": "doc_000237", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Laurel-11, Cedar-3B produced 0.726 macro F1 on NereidNotes-3 for Project Nereid on 2026-04-07."}, {"sent_id": 3, "text": "The 2026-05-23 result for Project Saffron put Mira-3B on MemoTrace-2 at 0.613 latency efficiency score, with Node Sycamore-13 handling the run."}, {"sent_id": 4, "text": "Project Sonata received 3 GPUs from Node Fir-10 in the 2026-06-06 allocation log."}, {"sent_id": 5, "text": "For evidence retrieval, the 2026-05-17 note says Project Meridian went with Atlas-3B."}, {"sent_id": 6, "text": "The 2026-04-24 compute note links Node Cedar-02 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 7, "text": "Project Aster had a failed run on 2026-05-15 involving River-8B and Node Rowan-09."}, {"sent_id": 8, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 9, "text": "Project Sonata logged a personnel update on 2026-06-23: Rohan Singh took the retrieval owner responsibilities."}, {"sent_id": 10, "text": "Project Aster logged a personnel update on 2026-05-04: Dr. Noah Vale took the lead responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 237"}
{"doc_id": "doc_000238", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid had a method update on 2026-04-11."}, {"sent_id": 3, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}, {"sent_id": 4, "text": "On 2026-06-19, Project Anchor shifted Milestone H1: 2026-07-13 became 2026-07-17."}, {"sent_id": 5, "text": "Project Saffron recorded a model choice for claim classification on 2026-05-24."}, {"sent_id": 6, "text": "The chosen model was Quartz-7B."}, {"sent_id": 7, "text": "The 2026-05-16 failure note says Helix-8B on Node Poplar-12 failed for Project Anchor because of a missing-index error."}, {"sent_id": 8, "text": "Project Meridian had a method update on 2026-06-20."}, {"sent_id": 9, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 10, "text": "Project Anchor had a failed run on 2026-06-12 involving Helix-8B and Node Poplar-12."}, {"sent_id": 11, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 12, "text": "Helix-8B was the model selected by Project Anchor on 2026-04-30 for the claim classification track."}], "source": "synthetic_rewritten", "title": "Project Update 238"}
{"doc_id": "doc_000239", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-22 failure note says River-7B on Node Rowan-09 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 3, "text": "A compute allocation for Project Sonata was logged on 2026-04-30."}, {"sent_id": 4, "text": "Node Pine-07 supplied 3 GPUs."}, {"sent_id": 5, "text": "The 2026-04-26 method entry for Project Aster moved from alias expansion to dense retrieval."}, {"sent_id": 6, "text": "Project Nereid logged a personnel update on 2026-06-09: Vera Kim took the lead responsibilities."}, {"sent_id": 7, "text": "The 2026-05-24 method entry for Project Nereid moved from document chunking to metric smoothing."}, {"sent_id": 8, "text": "Lumen-7B was the model selected by Project Anchor on 2026-06-17 for the calibration track."}, {"sent_id": 9, "text": "Project Sonata logged a personnel update on 2026-06-02: Dr. Jonas Rios took the lead responsibilities."}, {"sent_id": 10, "text": "Project Anchor's Milestone Z1 deadline change on 2026-06-04 replaced 2026-06-12 with 2026-06-16."}], "source": "synthetic_rewritten", "title": "Lab Digest 239"}
{"doc_id": "doc_000240", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For claim classification, the 2026-04-16 note says Project Anchor went with Helix-7B."}, {"sent_id": 3, "text": "The 2026-06-15 result for Project Nereid put Atlas-3B on RiverBench-2 at 0.654 evidence F1, with Node Hazel-14 handling the run."}, {"sent_id": 4, "text": "On 2026-06-28, Project Anchor retired metric smoothing in favor of hard-negative mining."}, {"sent_id": 5, "text": "On 2026-04-23, Project Meridian shifted Milestone B2: 2026-05-07 became 2026-05-13."}, {"sent_id": 6, "text": "Project Anchor ran Helix-8B on CedarQA-3 using Node Poplar-12 on 2026-06-15."}, {"sent_id": 7, "text": "The reported latency efficiency score was 0.797."}, {"sent_id": 8, "text": "The 2026-05-20 method entry for Project Anchor moved from confidence calibration to chain verification."}, {"sent_id": 9, "text": "On 2026-05-27, Project Meridian shifted Milestone P1: 2026-06-04 became 2026-06-12."}], "source": "synthetic_rewritten", "title": "Review Packet 240"}
{"doc_id": "doc_000241", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-05-27, Project Saffron shifted Milestone L1: 2026-06-22 became 2026-07-02."}, {"sent_id": 3, "text": "Project Nereid ran Nova-3B on LabQA using Node Aspen-01 on 2026-05-25."}, {"sent_id": 4, "text": "The reported macro F1 was 0.720."}, {"sent_id": 5, "text": "Using Node Elm-08, Nimbus-3B produced 0.723 latency efficiency score on MemoTrace for Project Meridian on 2026-04-18."}, {"sent_id": 6, "text": "On 2026-04-15, Project Saffron retired hybrid retrieval in favor of document chunking."}, {"sent_id": 7, "text": "On 2026-04-30, Project Sonata shifted Milestone D2: 2026-05-15 became 2026-05-11."}, {"sent_id": 8, "text": "A compute allocation for Project Sonata was logged on 2026-05-30."}, {"sent_id": 9, "text": "Node Maple-01 supplied 1 GPU."}, {"sent_id": 10, "text": "Project Saffron recorded a model choice for reranking on 2026-05-17."}, {"sent_id": 11, "text": "The chosen model was Vela-7B."}], "source": "synthetic_rewritten", "title": "Run Brief 241"}
{"doc_id": "doc_000242", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-05-10, Project Aster retired contrastive tuning in favor of threshold search."}, {"sent_id": 3, "text": "The staffing note for Project Meridian was dated 2026-04-27."}, {"sent_id": 4, "text": "It named Nadia Chen for the evaluation owner duties."}, {"sent_id": 5, "text": "The 2026-06-26 compute note links Node Laurel-11 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 6, "text": "The 2026-04-03 compute note links Node Poplar-12 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 7, "text": "For calibration, the 2026-05-13 note says Project Saffron went with Mira-8B."}, {"sent_id": 8, "text": "The 2026-04-03 failure note says Nova-8B on Node Aspen-01 failed for Project Meridian because of a missing-index error."}, {"sent_id": 9, "text": "In the 2026-06-22 staffing section, Project Saffron listed Lena Sol as the evaluation owner."}], "source": "synthetic_rewritten", "title": "Review Packet 242"}
{"doc_id": "doc_000243", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-26, Project Aster retired LoRA adaptation in favor of alias expansion."}, {"sent_id": 3, "text": "Milestone L1 for Project Sonata was revised on 2026-06-23."}, {"sent_id": 4, "text": "The schedule moved from 2026-07-04 to 2026-06-29."}, {"sent_id": 5, "text": "In the 2026-04-28 staffing section, Project Meridian listed Elian Ames as the evaluation owner."}, {"sent_id": 6, "text": "Vela-7B was the model selected by Project Saffron on 2026-06-04 for the reranking track."}, {"sent_id": 7, "text": "Milestone B2 for Project Nereid was revised on 2026-05-22."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-31 to 2026-06-05."}, {"sent_id": 9, "text": "Project Meridian had a failed run on 2026-05-07 involving Atlas-7B and Node Hazel-14."}, {"sent_id": 10, "text": "The recorded cause was a missing-index error."}, {"sent_id": 11, "text": "The 2026-06-21 result for Project Saffron put Vela-3B on LumenFacts-3 at 0.715 accuracy, with Node Fir-10 handling the run."}], "source": "synthetic_rewritten", "title": "Run Brief 243"}
{"doc_id": "doc_000244", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian ran Cedar-8B on VestaLogs-2 using Node Laurel-11 on 2026-04-19."}, {"sent_id": 3, "text": "The reported accuracy was 0.861."}, {"sent_id": 4, "text": "The 2026-06-07 method entry for Project Meridian moved from threshold search to reward reranking."}, {"sent_id": 5, "text": "Project Anchor ran Finch-3B on VestaLogs using Node Juniper-06 on 2026-05-05."}, {"sent_id": 6, "text": "The reported latency efficiency score was 0.521."}, {"sent_id": 7, "text": "Project Meridian logged a personnel update on 2026-06-23: Nadia Singh took the data steward responsibilities."}, {"sent_id": 8, "text": "The 2026-05-20 compute note links Node Spruce-03 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 9, "text": "On 2026-04-22, Project Meridian retired structured prompting in favor of QLoRA adaptation."}, {"sent_id": 10, "text": "Project Aster received 1 GPU from Node Poplar-12 in the 2026-04-01 allocation log."}, {"sent_id": 11, "text": "Quartz-3B was the model selected by Project Saffron on 2026-06-12 for the claim classification track."}, {"sent_id": 12, "text": "Project Saffron logged a personnel update on 2026-06-21: Priya Moss took the data steward responsibilities."}], "source": "synthetic_rewritten", "title": "Review Packet 244"}
{"doc_id": "doc_000245", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Sonata ran Vela-3B on LabQA-2 using Node Fir-10 on 2026-06-22."}, {"sent_id": 3, "text": "The reported evidence F1 was 0.580."}, {"sent_id": 4, "text": "Project Sonata received 1 GPU from Node Fir-10 in the 2026-05-06 allocation log."}, {"sent_id": 5, "text": "On 2026-05-22, Project Saffron shifted Milestone R1: 2026-06-05 became 2026-06-11."}, {"sent_id": 6, "text": "Project Anchor recorded a model choice for claim classification on 2026-04-22."}, {"sent_id": 7, "text": "The chosen model was Helix-8B."}, {"sent_id": 8, "text": "Project Nereid ran Atlas-7B on RiverBench-2 using Node Hazel-14 on 2026-05-02."}, {"sent_id": 9, "text": "The reported macro F1 was 0.744."}, {"sent_id": 10, "text": "The 2026-05-16 compute note links Node Aspen-01 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 11, "text": "For Project Aster, the run using Lumen-7B on Node Spruce-03 did not complete on 2026-04-03; the issue was an out-of-memory error."}, {"sent_id": 12, "text": "The 2026-05-20 compute note links Node Maple-01 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 13, "text": "Lumen-3B was the model selected by Project Anchor on 2026-06-30 for the calibration track."}], "source": "synthetic_rewritten", "title": "Run Brief 245"}
{"doc_id": "doc_000246", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian had a failed run on 2026-04-03 involving Nimbus-3B and Node Elm-08."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "Nested Verifier Study 4 appeared in a 2026-04-24 review entry using rank fusion; the note said it did not use a reward model."}, {"sent_id": 5, "text": "On 2026-04-19, Project Sonata retired calibrated voting in favor of confidence calibration."}, {"sent_id": 6, "text": "Project Anchor ran Lumen-8B on MemoTrace-3 using Node Spruce-03 on 2026-06-06."}, {"sent_id": 7, "text": "The reported accuracy was 0.803."}, {"sent_id": 8, "text": "The 2026-06-25 failure note says Finch-8B on Node Juniper-06 failed for Project Aster because of an unstable-validation-loss error."}, {"sent_id": 9, "text": "Milestone B2 for Project Meridian was revised on 2026-04-17."}, {"sent_id": 10, "text": "The schedule moved from 2026-05-01 to 2026-05-11."}, {"sent_id": 11, "text": "A compute allocation for Project Nereid was logged on 2026-04-04."}, {"sent_id": 12, "text": "Node Hazel-14 supplied 1 GPU."}, {"sent_id": 13, "text": "Project Meridian's Milestone J1 deadline change on 2026-04-08 replaced 2026-04-28 with 2026-05-06."}], "source": "synthetic_rewritten", "title": "Review Packet 246"}
{"doc_id": "doc_000247", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-06-30 staffing section, Project Nereid listed Darian Hale as the retrieval owner."}, {"sent_id": 3, "text": "Orchid-8B was the model selected by Project Anchor on 2026-04-14 for the reranking track."}, {"sent_id": 4, "text": "Using Node Birch-04, Kestrel-3B produced 0.672 macro F1 on OrionBench-3 for Project Nereid on 2026-06-01."}, {"sent_id": 5, "text": "On 2026-06-17, Project Saffron shifted Milestone F1: 2026-06-25 became 2026-07-03."}, {"sent_id": 6, "text": "Milestone B2 for Project Meridian was revised on 2026-04-24."}, {"sent_id": 7, "text": "The schedule moved from 2026-05-18 to 2026-05-26."}, {"sent_id": 8, "text": "Project Nereid's Milestone B2 deadline change on 2026-06-05 replaced 2026-06-14 with 2026-06-23."}, {"sent_id": 9, "text": "Project Aster's Milestone B1 deadline change on 2026-05-29 replaced 2026-06-19 with 2026-06-28."}], "source": "synthetic_rewritten", "title": "Run Brief 247"}
{"doc_id": "doc_000248", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron ran Marble-8B on VestaLogs-3 using Node Maple-01 on 2026-04-07."}, {"sent_id": 3, "text": "The reported accuracy was 0.661."}, {"sent_id": 4, "text": "For Project Saffron, the run using Mira-7B on Node Sycamore-13 did not complete on 2026-04-16; the issue was a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-06-22 result for Project Anchor put Orchid-8B on LumenFacts at 0.515 latency efficiency score, with Node Cedar-02 handling the run."}, {"sent_id": 6, "text": "On 2026-04-08, Project Anchor retired temporal filtering in favor of alias expansion."}, {"sent_id": 7, "text": "The 2026-04-19 method entry for Project Nereid moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 8, "text": "For error analysis, the 2026-04-03 note says Project Meridian went with Cedar-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 248"}
{"doc_id": "doc_000249", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata received 3 GPUs from Node Sycamore-13 in the 2026-05-16 allocation log."}, {"sent_id": 3, "text": "A compute allocation for Project Sonata was logged on 2026-05-01."}, {"sent_id": 4, "text": "Node Pine-07 supplied 1 GPU."}, {"sent_id": 5, "text": "The staffing note for Project Aster was dated 2026-04-07."}, {"sent_id": 6, "text": "It named Felix Brooks for the retrieval owner duties."}, {"sent_id": 7, "text": "In the 2026-04-06 staffing section, Project Meridian listed Camila Brooks as the data steward."}, {"sent_id": 8, "text": "For calibration, the 2026-05-08 note says Project Aster went with Lumen-8B."}, {"sent_id": 9, "text": "On 2026-05-12, Project Anchor shifted Milestone H1: 2026-06-05 became 2026-06-13."}, {"sent_id": 10, "text": "For reranking, the 2026-05-27 note says Project Sonata went with Vela-8B."}, {"sent_id": 11, "text": "The 2026-04-19 method entry for Project Aster moved from BM25 retrieval to evidence pooling."}, {"sent_id": 12, "text": "The review note for Pale Compass Study 4 was dated 2026-06-01."}, {"sent_id": 13, "text": "It described the method as data mixing and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Run Brief 249"}
{"doc_id": "doc_000250", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For claim classification, the 2026-06-21 note says Project Nereid went with Kestrel-7B."}, {"sent_id": 3, "text": "The 2026-04-11 compute note links Node Fir-10 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 4, "text": "Project Nereid's Milestone B2 deadline change on 2026-05-29 replaced 2026-06-17 with 2026-06-14."}, {"sent_id": 5, "text": "Milestone J1 for Project Nereid was revised on 2026-06-25."}, {"sent_id": 6, "text": "The schedule moved from 2026-07-10 to 2026-07-06."}, {"sent_id": 7, "text": "The 2026-04-18 compute note links Node Maple-01 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 8, "text": "Project Sonata had a method update on 2026-05-03."}, {"sent_id": 9, "text": "The old approach was query rewriting, and the replacement was sentence pruning."}, {"sent_id": 10, "text": "For reranking, the 2026-06-28 note says Project Meridian went with Nova-7B."}, {"sent_id": 11, "text": "The 2026-06-11 compute note links Node Aspen-01 to Project Nereid with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Lab Digest 250"}
{"doc_id": "doc_000251", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-06-28, Project Anchor retired LoRA adaptation in favor of alias expansion."}, {"sent_id": 3, "text": "In the 2026-05-12 staffing section, Project Saffron listed Petra Adler as the data steward."}, {"sent_id": 4, "text": "The staffing note for Project Sonata was dated 2026-05-24."}, {"sent_id": 5, "text": "It named Dr. Jonas Rios for the lead duties."}, {"sent_id": 6, "text": "Project Meridian ran Atlas-8B on SignalSet-3 using Node Hazel-14 on 2026-05-30."}, {"sent_id": 7, "text": "The reported accuracy was 0.669."}, {"sent_id": 8, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-05-27."}, {"sent_id": 9, "text": "The chosen model was Aster-3B."}, {"sent_id": 10, "text": "The staffing note for Project Meridian was dated 2026-04-27."}, {"sent_id": 11, "text": "It named Camila Quinn for the evaluation owner duties."}, {"sent_id": 12, "text": "In the 2026-06-30 staffing section, Project Saffron listed Lena Sol as the evaluation owner."}, {"sent_id": 13, "text": "Project Aster had a method update on 2026-04-09."}, {"sent_id": 14, "text": "The old approach was threshold search, and the replacement was reward reranking."}], "source": "synthetic_rewritten", "title": "Project Update 251"}
{"doc_id": "doc_000252", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-22 method entry for Project Nereid moved from late interaction to BM25 retrieval."}, {"sent_id": 3, "text": "Milestone L1 for Project Sonata was revised on 2026-06-24."}, {"sent_id": 4, "text": "The schedule moved from 2026-07-05 to 2026-06-30."}, {"sent_id": 5, "text": "Project Aster had a method update on 2026-05-27."}, {"sent_id": 6, "text": "The old approach was reward reranking, and the replacement was data mixing."}, {"sent_id": 7, "text": "The 2026-05-13 method entry for Project Anchor moved from late interaction to BM25 retrieval."}, {"sent_id": 8, "text": "The 2026-06-23 result for Project Nereid put Atlas-7B on RiverBench-2 at 0.744 evidence F1, with Node Hazel-14 handling the run."}, {"sent_id": 9, "text": "Project Nereid had a method update on 2026-06-14."}, {"sent_id": 10, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 11, "text": "Project Nereid's Milestone D1 deadline change on 2026-05-08 replaced 2026-05-25 with 2026-05-30."}, {"sent_id": 12, "text": "The staffing note for Project Aster was dated 2026-05-05."}, {"sent_id": 13, "text": "It named Dr. Noah Chen for the retrieval owner duties."}, {"sent_id": 14, "text": "Project Aster had a method update on 2026-04-29."}, {"sent_id": 15, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}], "source": "synthetic_rewritten", "title": "Lab Digest 252"}
{"doc_id": "doc_000253", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-04-28 result for Project Saffron put Aster-8B on SignalSet at 0.829 latency efficiency score, with Node Willow-05 handling the run."}, {"sent_id": 3, "text": "Project Anchor logged a personnel update on 2026-05-11: Dr. Arun Bauer took the evaluation owner responsibilities."}, {"sent_id": 4, "text": "The 2026-04-30 failure note says Aster-3B on Node Willow-05 failed for Project Saffron because of a missing-index error."}, {"sent_id": 5, "text": "In the 2026-06-22 staffing section, Project Sonata listed Rohan Park as the lead."}, {"sent_id": 6, "text": "Project Meridian ran Nimbus-7B on MemoTrace using Node Elm-08 on 2026-06-23."}, {"sent_id": 7, "text": "The reported accuracy was 0.783."}, {"sent_id": 8, "text": "The 2026-05-28 failure note says Nova-3B on Node Aspen-01 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 9, "text": "Milestone F1 for Project Sonata was revised on 2026-06-04."}, {"sent_id": 10, "text": "The schedule moved from 2026-06-27 to 2026-06-25."}, {"sent_id": 11, "text": "The 2026-06-23 result for Project Nereid put Kestrel-8B on OrionBench-3 at 0.822 macro F1, with Node Birch-04 handling the run."}, {"sent_id": 12, "text": "Project Aster ran River-3B on RiverBench using Node Rowan-09 on 2026-06-13."}, {"sent_id": 13, "text": "The reported macro F1 was 0.794."}], "source": "synthetic_rewritten", "title": "Project Update 253"}
{"doc_id": "doc_000254", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Spruce-03, Lumen-7B produced 0.848 evidence F1 on TraceEval for Project Aster on 2026-04-06."}, {"sent_id": 3, "text": "Project Sonata's Milestone X1 deadline change on 2026-04-13 replaced 2026-04-30 with 2026-05-09."}, {"sent_id": 4, "text": "Project Meridian ran Nimbus-8B on MemoTrace using Node Elm-08 on 2026-06-16."}, {"sent_id": 5, "text": "The reported latency efficiency score was 0.753."}, {"sent_id": 6, "text": "Project Saffron's Milestone L1 deadline change on 2026-04-03 replaced 2026-04-29 with 2026-05-05."}, {"sent_id": 7, "text": "In the 2026-06-01 staffing section, Project Meridian listed Elian Shah as the data steward."}, {"sent_id": 8, "text": "Using Node Elm-08, Nimbus-3B produced 0.723 accuracy on MemoTrace for Project Meridian on 2026-06-09."}, {"sent_id": 9, "text": "On 2026-04-15, Project Sonata retired reward reranking in favor of data mixing."}], "source": "synthetic_rewritten", "title": "Lab Digest 254"}
{"doc_id": "doc_000255", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-09 result for Project Sonata put Quartz-8B on OrionBench at 0.592 evidence F1, with Node Pine-07 handling the run."}, {"sent_id": 3, "text": "The 2026-06-14 method entry for Project Anchor moved from hard-negative mining to LoRA adaptation."}, {"sent_id": 4, "text": "A compute allocation for Project Meridian was logged on 2026-06-20."}, {"sent_id": 5, "text": "Node Elm-08 supplied 4 GPUs."}, {"sent_id": 6, "text": "Aster-3B was the model selected by Project Sonata on 2026-05-28 for the evidence retrieval track."}, {"sent_id": 7, "text": "Project Aster logged a personnel update on 2026-06-16: Anika Sato took the lead responsibilities."}, {"sent_id": 8, "text": "Project Saffron's Milestone D2 deadline change on 2026-04-20 replaced 2026-05-10 with 2026-05-18."}, {"sent_id": 9, "text": "Project Sonata's Milestone X1 deadline change on 2026-04-29 replaced 2026-05-06 with 2026-05-05."}, {"sent_id": 10, "text": "For evidence retrieval, the 2026-04-14 note says Project Saffron went with Aster-7B."}, {"sent_id": 11, "text": "The review note for Amber Ranking Study 2 was dated 2026-05-08."}, {"sent_id": 12, "text": "It described the method as confidence calibration and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Project Update 255"}
{"doc_id": "doc_000256", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-16 compute note links Node Pine-07 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 3, "text": "The 2026-04-25 result for Project Meridian put Nova-7B on LumenFacts-2 at 0.675 latency efficiency score, with Node Aspen-01 handling the run."}, {"sent_id": 4, "text": "Project Anchor logged a personnel update on 2026-05-05: Leo Hale took the data steward responsibilities."}, {"sent_id": 5, "text": "The 2026-06-30 result for Project Nereid put Atlas-3B on RiverBench-2 at 0.774 macro F1, with Node Hazel-14 handling the run."}, {"sent_id": 6, "text": "Project Sonata had a method update on 2026-06-07."}, {"sent_id": 7, "text": "The old approach was dense retrieval, and the replacement was rank fusion."}, {"sent_id": 8, "text": "Project Nereid logged a personnel update on 2026-06-16: Vera Torres took the retrieval owner responsibilities."}, {"sent_id": 9, "text": "Project Saffron logged a personnel update on 2026-05-18: Lena Sol took the evaluation owner responsibilities."}, {"sent_id": 10, "text": "Project Nereid logged a personnel update on 2026-06-15: Evan Iyer took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 256"}
{"doc_id": "doc_000257", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor received 2 GPUs from Node Poplar-12 in the 2026-05-23 allocation log."}, {"sent_id": 3, "text": "Project Saffron had a failed run on 2026-04-09 involving Quartz-7B and Node Pine-07."}, {"sent_id": 4, "text": "The recorded cause was a missing-index error."}, {"sent_id": 5, "text": "Project Anchor logged a personnel update on 2026-05-26: Iris Stone took the data steward responsibilities."}, {"sent_id": 6, "text": "Project Saffron received 2 GPUs from Node Maple-01 in the 2026-04-24 allocation log."}, {"sent_id": 7, "text": "Project Meridian had a failed run on 2026-06-18 involving Kestrel-3B and Node Birch-04."}, {"sent_id": 8, "text": "The recorded cause was a missing-index error."}, {"sent_id": 9, "text": "The 2026-06-22 result for Project Aster put Lumen-3B on TraceEval at 0.788 evidence F1, with Node Spruce-03 handling the run."}], "source": "synthetic_rewritten", "title": "Project Update 257"}
{"doc_id": "doc_000258", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-20 compute note links Node Rowan-09 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 3, "text": "Project Saffron ran Aster-7B on SignalSet using Node Willow-05 on 2026-06-16."}, {"sent_id": 4, "text": "The reported accuracy was 0.649."}, {"sent_id": 5, "text": "Project Anchor recorded a model choice for claim classification on 2026-05-20."}, {"sent_id": 6, "text": "The chosen model was Helix-7B."}, {"sent_id": 7, "text": "Project Anchor had a method update on 2026-06-14."}, {"sent_id": 8, "text": "The old approach was confidence calibration, and the replacement was chain verification."}, {"sent_id": 9, "text": "Milestone B1 for Project Anchor was revised on 2026-04-16."}, {"sent_id": 10, "text": "The schedule moved from 2026-05-12 to 2026-05-18."}, {"sent_id": 11, "text": "Project Anchor had a method update on 2026-04-05."}, {"sent_id": 12, "text": "The old approach was hard-negative mining, and the replacement was LoRA adaptation."}, {"sent_id": 13, "text": "Project Aster had a method update on 2026-06-07."}, {"sent_id": 14, "text": "The old approach was chain verification, and the replacement was structured prompting."}, {"sent_id": 15, "text": "A compute allocation for Project Aster was logged on 2026-05-04."}, {"sent_id": 16, "text": "Node Juniper-06 supplied 3 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 258"}
{"doc_id": "doc_000259", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Using Node Cedar-02, Orchid-3B produced 0.860 evidence F1 on LabQA-3 for Project Aster on 2026-05-23."}, {"sent_id": 3, "text": "On 2026-05-28, Project Nereid retired calibrated voting in favor of confidence calibration."}, {"sent_id": 4, "text": "The 2026-04-04 compute note links Node Cedar-02 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 5, "text": "Project Aster had a method update on 2026-06-07."}, {"sent_id": 6, "text": "The old approach was confidence calibration, and the replacement was chain verification."}, {"sent_id": 7, "text": "The review note for Amber Ranking Study 3 was dated 2026-05-15."}, {"sent_id": 8, "text": "It described the method as hybrid retrieval and said it did not use a reward model."}, {"sent_id": 9, "text": "The 2026-06-20 compute note links Node Poplar-12 to Project Anchor with an allocation of 2 GPUs."}], "source": "synthetic_rewritten", "title": "Project Update 259"}
{"doc_id": "doc_000260", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-08, Project Anchor shifted Milestone N1: 2026-04-30 became 2026-05-06."}, {"sent_id": 3, "text": "Atlas-8B was the model selected by Project Nereid on 2026-05-07 for the evidence retrieval track."}, {"sent_id": 4, "text": "For claim classification, the 2026-05-03 note says Project Aster went with Helix-7B."}, {"sent_id": 5, "text": "Project Anchor received 4 GPUs from Node Poplar-12 in the 2026-06-27 allocation log."}, {"sent_id": 6, "text": "The 2026-05-24 method entry for Project Sonata moved from temporal filtering to alias expansion."}, {"sent_id": 7, "text": "The staffing note for Project Meridian was dated 2026-06-09."}, {"sent_id": 8, "text": "It named Elian Shah for the data steward duties."}, {"sent_id": 9, "text": "For error analysis, the 2026-06-21 note says Project Sonata went with Marble-8B."}, {"sent_id": 10, "text": "On 2026-05-21, Project Meridian shifted Milestone V1: 2026-06-16 became 2026-06-26."}], "source": "synthetic_rewritten", "title": "Run Brief 260"}
{"doc_id": "doc_000261", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-06-30 staffing section, Project Nereid listed Vera Kim as the lead."}, {"sent_id": 3, "text": "The 2026-05-01 compute note links Node Birch-04 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 4, "text": "The 2026-04-26 result for Project Nereid put Kestrel-3B on OrionBench-3 at 0.852 macro F1, with Node Birch-04 handling the run."}, {"sent_id": 5, "text": "The 2026-06-25 failure note says Marble-8B on Node Maple-01 failed for Project Saffron because of a checkpoint-mismatch error."}, {"sent_id": 6, "text": "Milestone L1 for Project Sonata was revised on 2026-04-30."}, {"sent_id": 7, "text": "The schedule moved from 2026-05-21 to 2026-05-30."}, {"sent_id": 8, "text": "The 2026-06-04 failure note says Cedar-7B on Node Laurel-11 failed for Project Meridian because of a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Review Packet 261"}
{"doc_id": "doc_000262", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-04 compute note links Node Elm-08 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "Helix-3B was the model selected by Project Aster on 2026-06-16 for the claim classification track."}, {"sent_id": 4, "text": "Milestone L1 for Project Saffron was revised on 2026-05-19."}, {"sent_id": 5, "text": "The schedule moved from 2026-06-04 to 2026-06-08."}, {"sent_id": 6, "text": "Milestone P1 for Project Nereid was revised on 2026-04-01."}, {"sent_id": 7, "text": "The schedule moved from 2026-04-24 to 2026-04-22."}, {"sent_id": 8, "text": "For reranking, the 2026-06-14 note says Project Saffron went with Vela-7B."}, {"sent_id": 9, "text": "A compute allocation for Project Saffron was logged on 2026-05-04."}, {"sent_id": 10, "text": "Node Willow-05 supplied 4 GPUs."}, {"sent_id": 11, "text": "On 2026-05-28, Project Meridian retired chain verification in favor of structured prompting."}], "source": "synthetic_rewritten", "title": "Run Brief 262"}
{"doc_id": "doc_000263", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone T1 for Project Aster was revised on 2026-04-06."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-01 to 2026-05-06."}, {"sent_id": 4, "text": "For calibration, the 2026-04-22 note says Project Anchor went with Lumen-8B."}, {"sent_id": 5, "text": "The staffing note for Project Saffron was dated 2026-06-01."}, {"sent_id": 6, "text": "It named Petra Gray for the evaluation owner duties."}, {"sent_id": 7, "text": "Milestone P1 for Project Meridian was revised on 2026-05-20."}, {"sent_id": 8, "text": "The schedule moved from 2026-06-07 to 2026-06-17."}, {"sent_id": 9, "text": "The staffing note for Project Anchor was dated 2026-05-05."}, {"sent_id": 10, "text": "It named Iris Stone for the data steward duties."}, {"sent_id": 11, "text": "On 2026-05-02, Project Saffron retired calibrated voting in favor of confidence calibration."}, {"sent_id": 12, "text": "Milestone J1 for Project Meridian was revised on 2026-05-01."}, {"sent_id": 13, "text": "The schedule moved from 2026-05-21 to 2026-05-29."}, {"sent_id": 14, "text": "The review note for Pale Compass Study 4 was dated 2026-04-29."}, {"sent_id": 15, "text": "It described the method as reward reranking and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Review Packet 263"}
{"doc_id": "doc_000264", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-04-19 method entry for Project Saffron moved from chain verification to structured prompting."}, {"sent_id": 3, "text": "The 2026-05-19 result for Project Aster put Lumen-7B on TraceEval at 0.698 evidence F1, with Node Spruce-03 handling the run."}, {"sent_id": 4, "text": "In the 2026-06-15 staffing section, Project Anchor listed Kira Frost as the data steward."}, {"sent_id": 5, "text": "On 2026-06-26, Project Meridian shifted Milestone P1: 2026-07-14 became 2026-07-20."}, {"sent_id": 6, "text": "For reranking, the 2026-04-24 note says Project Meridian went with Nova-8B."}, {"sent_id": 7, "text": "For error analysis, the 2026-06-25 note says Project Sonata went with Marble-3B."}, {"sent_id": 8, "text": "Project Meridian had a failed run on 2026-05-21 involving Nova-3B and Node Aspen-01."}, {"sent_id": 9, "text": "The recorded cause was a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Run Brief 264"}
{"doc_id": "doc_000265", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron received 4 GPUs from Node Willow-05 in the 2026-04-10 allocation log."}, {"sent_id": 3, "text": "On 2026-06-11, Project Anchor shifted Milestone H1: 2026-07-05 became 2026-07-09."}, {"sent_id": 4, "text": "Project Nereid logged a personnel update on 2026-06-29: Vera Torres took the retrieval owner responsibilities."}, {"sent_id": 5, "text": "Milestone T1 for Project Anchor was revised on 2026-05-25."}, {"sent_id": 6, "text": "The schedule moved from 2026-06-14 to 2026-06-18."}, {"sent_id": 7, "text": "The 2026-05-06 compute note links Node Willow-05 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 8, "text": "On 2026-05-31, Project Sonata retired data mixing in favor of contrastive tuning."}, {"sent_id": 9, "text": "A compute allocation for Project Anchor was logged on 2026-05-18."}, {"sent_id": 10, "text": "Node Poplar-12 supplied 4 GPUs."}, {"sent_id": 11, "text": "The 2026-04-12 method entry for Project Saffron moved from metric smoothing to hard-negative mining."}, {"sent_id": 12, "text": "A compute allocation for Project Nereid was logged on 2026-06-13."}, {"sent_id": 13, "text": "Node Birch-04 supplied 1 GPU."}], "source": "synthetic_rewritten", "title": "Review Packet 265"}
{"doc_id": "doc_000266", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian had a method update on 2026-04-30."}, {"sent_id": 3, "text": "The old approach was late interaction, and the replacement was BM25 retrieval."}, {"sent_id": 4, "text": "Project Nereid received 1 GPU from Node Laurel-11 in the 2026-06-03 allocation log."}, {"sent_id": 5, "text": "Project Nereid had a method update on 2026-06-03."}, {"sent_id": 6, "text": "The old approach was sentence pruning, and the replacement was cross-encoder reranking."}, {"sent_id": 7, "text": "Project Saffron's Milestone X1 deadline change on 2026-04-09 replaced 2026-04-21 with 2026-04-25."}, {"sent_id": 8, "text": "Project Sonata had a method update on 2026-04-05."}, {"sent_id": 9, "text": "The old approach was confidence calibration, and the replacement was chain verification."}, {"sent_id": 10, "text": "For Project Sonata, the run using Marble-7B on Node Maple-01 did not complete on 2026-05-28; the issue was an out-of-memory error."}, {"sent_id": 11, "text": "Project Sonata had a method update on 2026-06-06."}, {"sent_id": 12, "text": "The old approach was hard-negative mining, and the replacement was LoRA adaptation."}, {"sent_id": 13, "text": "Project Saffron had a failed run on 2026-05-22 involving Vela-8B and Node Fir-10."}, {"sent_id": 14, "text": "The recorded cause was a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 266"}
{"doc_id": "doc_000267", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor had a failed run on 2026-05-14 involving Lumen-7B and Node Spruce-03."}, {"sent_id": 3, "text": "The recorded cause was a missing-index error."}, {"sent_id": 4, "text": "The 2026-06-05 failure note says Finch-7B on Node Juniper-06 failed for Project Aster because of an unstable-validation-loss error."}, {"sent_id": 5, "text": "Project Nereid ran Nova-8B on LabQA using Node Aspen-01 on 2026-04-14."}, {"sent_id": 6, "text": "The reported evidence F1 was 0.600."}, {"sent_id": 7, "text": "The 2026-05-20 method entry for Project Saffron moved from dense retrieval to rank fusion."}, {"sent_id": 8, "text": "The staffing note for Project Meridian was dated 2026-05-11."}, {"sent_id": 9, "text": "It named Elian Shah for the data steward duties."}, {"sent_id": 10, "text": "On 2026-06-13, Project Sonata retired cross-encoder reranking in favor of late interaction."}, {"sent_id": 11, "text": "Project Saffron recorded a model choice for claim classification on 2026-06-11."}, {"sent_id": 12, "text": "The chosen model was Quartz-7B."}, {"sent_id": 13, "text": "Project Sonata recorded a model choice for evidence retrieval on 2026-06-02."}, {"sent_id": 14, "text": "The chosen model was Aster-8B."}, {"sent_id": 15, "text": "Project Nereid had a method update on 2026-04-26."}, {"sent_id": 16, "text": "The old approach was dense retrieval, and the replacement was rank fusion."}], "source": "synthetic_rewritten", "title": "Review Packet 267"}
{"doc_id": "doc_000268", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Milestone D2 for Project Sonata was revised on 2026-06-08."}, {"sent_id": 3, "text": "The schedule moved from 2026-06-23 to 2026-06-19."}, {"sent_id": 4, "text": "Project Saffron logged a personnel update on 2026-06-29: Dr. Theo Grant took the data steward responsibilities."}, {"sent_id": 5, "text": "A compute allocation for Project Meridian was logged on 2026-06-26."}, {"sent_id": 6, "text": "Node Hazel-14 supplied 2 GPUs."}, {"sent_id": 7, "text": "The staffing note for Project Sonata was dated 2026-06-01."}, {"sent_id": 8, "text": "It named Dr. Jonas Nolan for the retrieval owner duties."}, {"sent_id": 9, "text": "Project Saffron recorded a model choice for calibration on 2026-05-26."}, {"sent_id": 10, "text": "The chosen model was Mira-7B."}, {"sent_id": 11, "text": "Project Sonata had a failed run on 2026-04-24 involving Quartz-8B and Node Pine-07."}, {"sent_id": 12, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 13, "text": "Project Meridian logged a personnel update on 2026-06-22: Elian Ames took the evaluation owner responsibilities."}], "source": "synthetic_rewritten", "title": "Run Brief 268"}
{"doc_id": "doc_000269", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The staffing note for Project Meridian was dated 2026-05-17."}, {"sent_id": 3, "text": "It named Mira Sato for the data steward duties."}, {"sent_id": 4, "text": "Project Sonata received 3 GPUs from Node Maple-01 in the 2026-06-11 allocation log."}, {"sent_id": 5, "text": "The 2026-05-05 result for Project Sonata put Marble-3B on NereidNotes at 0.586 evidence F1, with Node Maple-01 handling the run."}, {"sent_id": 6, "text": "Using Node Sycamore-13, Mira-8B produced 0.613 latency efficiency score on MemoTrace-2 for Project Saffron on 2026-04-27."}, {"sent_id": 7, "text": "On 2026-04-15, Project Meridian shifted Milestone J1: 2026-04-25 became 2026-05-01."}, {"sent_id": 8, "text": "Project Meridian received 4 GPUs from Node Aspen-01 in the 2026-05-20 allocation log."}], "source": "synthetic_rewritten", "title": "Review Packet 269"}
{"doc_id": "doc_000270", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For Project Sonata, the run using Vela-8B on Node Fir-10 did not complete on 2026-04-02; the issue was an unstable-validation-loss error."}, {"sent_id": 3, "text": "On 2026-04-15, Project Saffron shifted Milestone R1: 2026-04-29 became 2026-05-09."}, {"sent_id": 4, "text": "Milestone H1 for Project Aster was revised on 2026-04-23."}, {"sent_id": 5, "text": "The schedule moved from 2026-05-02 to 2026-05-07."}, {"sent_id": 6, "text": "Project Nereid's Milestone P1 deadline change on 2026-05-22 replaced 2026-06-14 with 2026-06-12."}, {"sent_id": 7, "text": "Project Nereid had a method update on 2026-04-04."}, {"sent_id": 8, "text": "The old approach was document chunking, and the replacement was metric smoothing."}, {"sent_id": 9, "text": "Orchid-3B was the model selected by Project Anchor on 2026-04-09 for the reranking track."}], "source": "synthetic_rewritten", "title": "Project Update 270"}
{"doc_id": "doc_000271", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For error analysis, the 2026-06-07 note says Project Meridian went with Cedar-7B."}, {"sent_id": 3, "text": "Project Anchor recorded a model choice for error analysis on 2026-06-11."}, {"sent_id": 4, "text": "The chosen model was Finch-8B."}, {"sent_id": 5, "text": "On 2026-05-06, Project Aster retired metric smoothing in favor of hard-negative mining."}, {"sent_id": 6, "text": "The 2026-06-20 method entry for Project Meridian moved from calibrated voting to confidence calibration."}, {"sent_id": 7, "text": "Project Aster recorded a model choice for evidence retrieval on 2026-06-17."}, {"sent_id": 8, "text": "The chosen model was River-3B."}, {"sent_id": 9, "text": "The 2026-05-07 failure note says Nimbus-7B on Node Elm-08 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 10, "text": "The 2026-05-30 compute note links Node Cedar-02 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 11, "text": "On 2026-06-19, Project Nereid shifted Milestone J1: 2026-07-04 became 2026-06-30."}], "source": "synthetic_rewritten", "title": "Lab Digest 271"}
{"doc_id": "doc_000272", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For evidence retrieval, the 2026-05-01 note says Project Anchor went with River-8B."}, {"sent_id": 3, "text": "A compute allocation for Project Anchor was logged on 2026-04-18."}, {"sent_id": 4, "text": "Node Juniper-06 supplied 2 GPUs."}, {"sent_id": 5, "text": "A compute allocation for Project Nereid was logged on 2026-06-20."}, {"sent_id": 6, "text": "Node Laurel-11 supplied 3 GPUs."}, {"sent_id": 7, "text": "A compute allocation for Project Aster was logged on 2026-05-16."}, {"sent_id": 8, "text": "Node Poplar-12 supplied 3 GPUs."}, {"sent_id": 9, "text": "For evidence retrieval, the 2026-06-05 note says Project Meridian went with Atlas-8B."}, {"sent_id": 10, "text": "On 2026-04-13, Project Saffron shifted Milestone R1: 2026-05-07 became 2026-05-15."}], "source": "synthetic_rewritten", "title": "Project Update 272"}
{"doc_id": "doc_000273", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata's Milestone R1 deadline change on 2026-05-25 replaced 2026-06-03 with 2026-06-12."}, {"sent_id": 3, "text": "For evidence retrieval, the 2026-04-19 note says Project Saffron went with Aster-8B."}, {"sent_id": 4, "text": "Project Anchor recorded a model choice for error analysis on 2026-06-09."}, {"sent_id": 5, "text": "The chosen model was Finch-8B."}, {"sent_id": 6, "text": "Project Aster had a failed run on 2026-05-15 involving Helix-7B and Node Poplar-12."}, {"sent_id": 7, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 8, "text": "For claim classification, the 2026-06-04 note says Project Aster went with Helix-8B."}, {"sent_id": 9, "text": "Nova-8B was the model selected by Project Meridian on 2026-05-15 for the reranking track."}, {"sent_id": 10, "text": "The 2026-06-12 failure note says Finch-3B on Node Juniper-06 failed for Project Aster because of an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Lab Digest 273"}
{"doc_id": "doc_000274", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata recorded a model choice for error analysis on 2026-05-03."}, {"sent_id": 3, "text": "The chosen model was Marble-8B."}, {"sent_id": 4, "text": "The 2026-04-21 result for Project Meridian put Kestrel-3B on CedarQA at 0.747 accuracy, with Node Birch-04 handling the run."}, {"sent_id": 5, "text": "A compute allocation for Project Sonata was logged on 2026-04-08."}, {"sent_id": 6, "text": "Node Maple-01 supplied 1 GPU."}, {"sent_id": 7, "text": "A compute allocation for Project Sonata was logged on 2026-04-18."}, {"sent_id": 8, "text": "Node Willow-05 supplied 3 GPUs."}, {"sent_id": 9, "text": "Using Node Maple-01, Marble-8B produced 0.616 macro F1 on NereidNotes for Project Sonata on 2026-05-12."}, {"sent_id": 10, "text": "For Project Nereid, the run using Kestrel-3B on Node Birch-04 did not complete on 2026-06-19; the issue was an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Project Update 274"}
{"doc_id": "doc_000275", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-09 compute note links Node Poplar-12 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 3, "text": "The 2026-05-08 compute note links Node Poplar-12 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 4, "text": "Atlas-3B was the model selected by Project Meridian on 2026-06-25 for the evidence retrieval track."}, {"sent_id": 5, "text": "The 2026-04-01 method entry for Project Saffron moved from calibrated voting to confidence calibration."}, {"sent_id": 6, "text": "The 2026-06-06 compute note links Node Maple-01 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 7, "text": "For error analysis, the 2026-05-20 note says Project Aster went with Finch-8B."}, {"sent_id": 8, "text": "A compute allocation for Project Saffron was logged on 2026-06-19."}, {"sent_id": 9, "text": "Node Pine-07 supplied 4 GPUs."}, {"sent_id": 10, "text": "Using Node Poplar-12, Helix-3B produced 0.887 latency efficiency score on CedarQA-3 for Project Anchor on 2026-06-23."}, {"sent_id": 11, "text": "Using Node Hazel-14, Atlas-7B produced 0.669 accuracy on SignalSet-3 for Project Meridian on 2026-05-04."}], "source": "synthetic_rewritten", "title": "Lab Digest 275"}
{"doc_id": "doc_000276", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Atlas-7B was the model selected by Project Nereid on 2026-05-27 for the evidence retrieval track."}, {"sent_id": 3, "text": "For claim classification, the 2026-06-14 note says Project Aster went with Helix-8B."}, {"sent_id": 4, "text": "On 2026-05-14, Project Anchor shifted Milestone N1: 2026-05-26 became 2026-06-03."}, {"sent_id": 5, "text": "The staffing note for Project Meridian was dated 2026-06-23."}, {"sent_id": 6, "text": "It named Mira Nolan for the evaluation owner duties."}, {"sent_id": 7, "text": "A compute allocation for Project Sonata was logged on 2026-05-27."}, {"sent_id": 8, "text": "Node Pine-07 supplied 1 GPU."}, {"sent_id": 9, "text": "The 2026-06-06 compute note links Node Hazel-14 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 10, "text": "On 2026-06-08, Project Aster shifted Milestone Z1: 2026-06-21 became 2026-06-30."}, {"sent_id": 11, "text": "Finch-3B was the model selected by Project Anchor on 2026-04-26 for the error analysis track."}, {"sent_id": 12, "text": "A compute allocation for Project Anchor was logged on 2026-04-25."}, {"sent_id": 13, "text": "Node Rowan-09 supplied 2 GPUs."}], "source": "synthetic_rewritten", "title": "Project Update 276"}
{"doc_id": "doc_000277", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Fir-10, Vela-3B produced 0.535 accuracy on LumenFacts-3 for Project Saffron on 2026-04-27."}, {"sent_id": 3, "text": "For evidence retrieval, the 2026-04-30 note says Project Anchor went with River-3B."}, {"sent_id": 4, "text": "The 2026-06-17 method entry for Project Anchor moved from evidence pooling to calibrated voting."}, {"sent_id": 5, "text": "A compute allocation for Project Sonata was logged on 2026-05-11."}, {"sent_id": 6, "text": "Node Fir-10 supplied 3 GPUs."}, {"sent_id": 7, "text": "Project Meridian had a method update on 2026-04-19."}, {"sent_id": 8, "text": "The old approach was late interaction, and the replacement was BM25 retrieval."}, {"sent_id": 9, "text": "In the 2026-05-26 staffing section, Project Saffron listed Dr. Theo Grant as the data steward."}, {"sent_id": 10, "text": "Project Sonata logged a personnel update on 2026-06-09: Samir Kwan took the retrieval owner responsibilities."}, {"sent_id": 11, "text": "Project Anchor had a method update on 2026-05-31."}, {"sent_id": 12, "text": "The old approach was chain verification, and the replacement was structured prompting."}, {"sent_id": 13, "text": "Using Node Fir-10, Vela-8B produced 0.745 latency efficiency score on LumenFacts-3 for Project Saffron on 2026-06-28."}], "source": "synthetic_rewritten", "title": "Lab Digest 277"}
{"doc_id": "doc_000278", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Sonata logged a personnel update on 2026-06-15: Sofia Frost took the lead responsibilities."}, {"sent_id": 3, "text": "The review note for Pale Compass Study 1 was dated 2026-05-06."}, {"sent_id": 4, "text": "It described the method as threshold search and said it did not use a reward model."}, {"sent_id": 5, "text": "Project Nereid had a method update on 2026-05-31."}, {"sent_id": 6, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 7, "text": "Project Saffron ran Marble-8B on VestaLogs-3 using Node Maple-01 on 2026-04-13."}, {"sent_id": 8, "text": "The reported accuracy was 0.631."}, {"sent_id": 9, "text": "Project Aster had a failed run on 2026-04-09 involving Lumen-7B and Node Spruce-03."}, {"sent_id": 10, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 11, "text": "Using Node Willow-05, Aster-3B produced 0.604 evidence F1 on RiverBench-3 for Project Sonata on 2026-05-04."}, {"sent_id": 12, "text": "Project Aster had a method update on 2026-06-14."}, {"sent_id": 13, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 14, "text": "Project Aster had a method update on 2026-06-03."}, {"sent_id": 15, "text": "The old approach was threshold search, and the replacement was reward reranking."}], "source": "synthetic_rewritten", "title": "Project Update 278"}
{"doc_id": "doc_000279", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Nereid had a method update on 2026-05-24."}, {"sent_id": 3, "text": "The old approach was confidence calibration, and the replacement was chain verification."}, {"sent_id": 4, "text": "On 2026-05-27, Project Nereid retired query rewriting in favor of sentence pruning."}, {"sent_id": 5, "text": "On 2026-05-22, Project Anchor shifted Milestone Z1: 2026-06-09 became 2026-06-19."}, {"sent_id": 6, "text": "Project Nereid's Milestone V1 deadline change on 2026-04-10 replaced 2026-04-21 with 2026-04-16."}, {"sent_id": 7, "text": "Project Nereid ran Cedar-7B on NereidNotes-3 using Node Laurel-11 on 2026-04-27."}, {"sent_id": 8, "text": "The reported macro F1 was 0.756."}, {"sent_id": 9, "text": "Project Sonata recorded a model choice for reranking on 2026-04-22."}, {"sent_id": 10, "text": "The chosen model was Vela-7B."}, {"sent_id": 11, "text": "Project Meridian recorded a model choice for reranking on 2026-04-19."}, {"sent_id": 12, "text": "The chosen model was Nova-7B."}, {"sent_id": 13, "text": "Using Node Pine-07, Quartz-8B produced 0.607 accuracy on CedarQA-2 for Project Saffron on 2026-06-01."}], "source": "synthetic_rewritten", "title": "Lab Digest 279"}
{"doc_id": "doc_000280", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Vela-8B was the model selected by Project Sonata on 2026-04-15 for the reranking track."}, {"sent_id": 3, "text": "The 2026-04-25 compute note links Node Juniper-06 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 4, "text": "The 2026-06-12 failure note says Kestrel-7B on Node Birch-04 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 5, "text": "Project Sonata had a failed run on 2026-04-23 involving Quartz-3B and Node Pine-07."}, {"sent_id": 6, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 7, "text": "The review note for Nested Verifier Study 2 was dated 2026-05-25."}, {"sent_id": 8, "text": "It described the method as hard-negative mining and said it did not use a reward model."}, {"sent_id": 9, "text": "Project Aster ran Lumen-7B on TraceEval using Node Spruce-03 on 2026-05-09."}, {"sent_id": 10, "text": "The reported macro F1 was 0.878."}, {"sent_id": 11, "text": "The 2026-05-14 failure note says Helix-7B on Node Poplar-12 failed for Project Anchor because of a missing-index error."}], "source": "synthetic_rewritten", "title": "Review Packet 280"}
{"doc_id": "doc_000281", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron's Milestone X1 deadline change on 2026-06-03 replaced 2026-06-25 with 2026-07-05."}, {"sent_id": 3, "text": "The 2026-05-30 failure note says Cedar-8B on Node Laurel-11 failed for Project Nereid because of an out-of-memory error."}, {"sent_id": 4, "text": "Project Aster's Milestone Z1 deadline change on 2026-04-09 replaced 2026-05-02 with 2026-04-30."}, {"sent_id": 5, "text": "Project Sonata ran Vela-8B on LabQA-2 using Node Fir-10 on 2026-05-02."}, {"sent_id": 6, "text": "The reported evidence F1 was 0.640."}, {"sent_id": 7, "text": "The 2026-04-04 method entry for Project Sonata moved from late interaction to BM25 retrieval."}, {"sent_id": 8, "text": "Project Meridian's Milestone D1 deadline change on 2026-05-20 replaced 2026-06-01 with 2026-06-09."}], "source": "synthetic_rewritten", "title": "Run Brief 281"}
{"doc_id": "doc_000282", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid logged a personnel update on 2026-05-19: Nora Bauer took the lead responsibilities."}, {"sent_id": 3, "text": "The 2026-05-17 result for Project Aster put Lumen-3B on TraceEval at 0.578 evidence F1, with Node Spruce-03 handling the run."}, {"sent_id": 4, "text": "Project Sonata had a method update on 2026-05-17."}, {"sent_id": 5, "text": "The old approach was threshold search, and the replacement was reward reranking."}, {"sent_id": 6, "text": "Project Aster recorded a model choice for error analysis on 2026-04-19."}, {"sent_id": 7, "text": "The chosen model was Finch-3B."}, {"sent_id": 8, "text": "Using Node Sycamore-13, Mira-3B produced 0.538 macro F1 on TraceEval-3 for Project Sonata on 2026-05-25."}, {"sent_id": 9, "text": "Kestrel-3B was the model selected by Project Nereid on 2026-04-10 for the claim classification track."}, {"sent_id": 10, "text": "The staffing note for Project Nereid was dated 2026-06-22."}, {"sent_id": 11, "text": "It named Evan Moss for the lead duties."}], "source": "synthetic_rewritten", "title": "Review Packet 282"}
{"doc_id": "doc_000283", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata recorded a model choice for claim classification on 2026-04-05."}, {"sent_id": 3, "text": "The chosen model was Quartz-3B."}, {"sent_id": 4, "text": "Project Saffron recorded a model choice for calibration on 2026-04-19."}, {"sent_id": 5, "text": "The chosen model was Mira-7B."}, {"sent_id": 6, "text": "Project Saffron recorded a model choice for claim classification on 2026-04-05."}, {"sent_id": 7, "text": "The chosen model was Quartz-8B."}, {"sent_id": 8, "text": "In the 2026-04-06 staffing section, Project Saffron listed Owen Marin as the data steward."}, {"sent_id": 9, "text": "Project Anchor received 2 GPUs from Node Poplar-12 in the 2026-06-06 allocation log."}, {"sent_id": 10, "text": "Project Sonata ran Marble-7B on NereidNotes using Node Maple-01 on 2026-05-25."}, {"sent_id": 11, "text": "The reported evidence F1 was 0.616."}], "source": "synthetic_rewritten", "title": "Run Brief 283"}
{"doc_id": "doc_000284", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron had a failed run on 2026-06-19 involving Marble-8B and Node Maple-01."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "Using Node Birch-04, Kestrel-3B produced 0.717 accuracy on CedarQA for Project Meridian on 2026-04-27."}, {"sent_id": 5, "text": "The 2026-06-06 result for Project Sonata put Marble-7B on NereidNotes at 0.556 evidence F1, with Node Maple-01 handling the run."}, {"sent_id": 6, "text": "Project Nereid logged a personnel update on 2026-06-01: Vera Kim took the lead responsibilities."}, {"sent_id": 7, "text": "The staffing note for Project Nereid was dated 2026-06-30."}, {"sent_id": 8, "text": "It named Evan Moss for the lead duties."}, {"sent_id": 9, "text": "Project Anchor had a method update on 2026-05-03."}, {"sent_id": 10, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 11, "text": "Project Saffron's Milestone R1 deadline change on 2026-04-22 replaced 2026-05-16 with 2026-05-24."}, {"sent_id": 12, "text": "River-7B was the model selected by Project Anchor on 2026-04-05 for the evidence retrieval track."}, {"sent_id": 13, "text": "In the 2026-06-30 staffing section, Project Anchor listed Selene Kim as the evaluation owner."}], "source": "synthetic_rewritten", "title": "Review Packet 284"}
{"doc_id": "doc_000285", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-29 failure note says Kestrel-3B on Node Birch-04 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 3, "text": "In the 2026-04-13 staffing section, Project Aster listed Talia Marin as the lead."}, {"sent_id": 4, "text": "On 2026-04-05, Project Aster retired query rewriting in favor of sentence pruning."}, {"sent_id": 5, "text": "The 2026-05-01 failure note says Nimbus-7B on Node Elm-08 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 6, "text": "The staffing note for Project Sonata was dated 2026-04-14."}, {"sent_id": 7, "text": "It named Dr. Jonas Nolan for the retrieval owner duties."}, {"sent_id": 8, "text": "The staffing note for Project Anchor was dated 2026-05-11."}, {"sent_id": 9, "text": "It named Leo Hale for the data steward duties."}, {"sent_id": 10, "text": "For evidence retrieval, the 2026-06-18 note says Project Aster went with River-7B."}], "source": "synthetic_rewritten", "title": "Run Brief 285"}
{"doc_id": "doc_000286", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-24, the Quiet Retriever Study 3 review reported evidence pooling and said it did not use a reward model."}, {"sent_id": 3, "text": "The 2026-06-30 result for Project Meridian put Nimbus-3B on MemoTrace at 0.813 latency efficiency score, with Node Elm-08 handling the run."}, {"sent_id": 4, "text": "Project Aster logged a personnel update on 2026-04-21: Talia Marin took the lead responsibilities."}, {"sent_id": 5, "text": "Finch-3B was the model selected by Project Anchor on 2026-06-02 for the error analysis track."}, {"sent_id": 6, "text": "Using Node Birch-04, Kestrel-7B produced 0.687 latency efficiency score on CedarQA for Project Meridian on 2026-04-20."}, {"sent_id": 7, "text": "Project Sonata's Milestone F1 deadline change on 2026-05-13 replaced 2026-05-26 with 2026-05-31."}, {"sent_id": 8, "text": "The staffing note for Project Meridian was dated 2026-05-19."}, {"sent_id": 9, "text": "It named Elian Shah for the data steward duties."}], "source": "synthetic_rewritten", "title": "Review Packet 286"}
{"doc_id": "doc_000287", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian had a failed run on 2026-05-08 involving Atlas-3B and Node Hazel-14."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "For Project Nereid, the run using Nova-7B on Node Aspen-01 did not complete on 2026-05-21; the issue was an out-of-memory error."}, {"sent_id": 5, "text": "The 2026-05-07 failure note says Lumen-8B on Node Spruce-03 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 6, "text": "For Project Aster, the run using Finch-7B on Node Juniper-06 did not complete on 2026-06-11; the issue was an unstable-validation-loss error."}, {"sent_id": 7, "text": "Project Meridian had a failed run on 2026-05-07 involving Nimbus-3B and Node Elm-08."}, {"sent_id": 8, "text": "The recorded cause was a missing-index error."}, {"sent_id": 9, "text": "Project Anchor had a method update on 2026-05-10."}, {"sent_id": 10, "text": "The old approach was sentence pruning, and the replacement was cross-encoder reranking."}, {"sent_id": 11, "text": "Project Nereid recorded a model choice for reranking on 2026-05-31."}, {"sent_id": 12, "text": "The chosen model was Nova-3B."}, {"sent_id": 13, "text": "Project Sonata recorded a model choice for calibration on 2026-04-14."}, {"sent_id": 14, "text": "The chosen model was Mira-8B."}, {"sent_id": 15, "text": "The 2026-05-31 method entry for Project Aster moved from calibrated voting to confidence calibration."}], "source": "synthetic_rewritten", "title": "Run Brief 287"}
{"doc_id": "doc_000288", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For evidence retrieval, the 2026-06-10 note says Project Meridian went with Atlas-7B."}, {"sent_id": 3, "text": "Project Anchor recorded a model choice for evidence retrieval on 2026-05-21."}, {"sent_id": 4, "text": "The chosen model was River-3B."}, {"sent_id": 5, "text": "The 2026-04-15 compute note links Node Rowan-09 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 6, "text": "A compute allocation for Project Meridian was logged on 2026-05-09."}, {"sent_id": 7, "text": "Node Birch-04 supplied 2 GPUs."}, {"sent_id": 8, "text": "Project Nereid had a method update on 2026-06-28."}, {"sent_id": 9, "text": "The old approach was LoRA adaptation, and the replacement was alias expansion."}, {"sent_id": 10, "text": "A compute allocation for Project Anchor was logged on 2026-06-13."}, {"sent_id": 11, "text": "Node Cedar-02 supplied 2 GPUs."}, {"sent_id": 12, "text": "On 2026-05-20, Project Aster shifted Milestone N1: 2026-05-27 became 2026-05-26."}, {"sent_id": 13, "text": "Finch-3B was the model selected by Project Anchor on 2026-06-04 for the error analysis track."}], "source": "synthetic_rewritten", "title": "Review Packet 288"}
{"doc_id": "doc_000289", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata had a failed run on 2026-05-30 involving Quartz-3B and Node Pine-07."}, {"sent_id": 3, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 4, "text": "Project Saffron ran Quartz-3B on CedarQA-2 using Node Pine-07 on 2026-06-09."}, {"sent_id": 5, "text": "The reported accuracy was 0.697."}, {"sent_id": 6, "text": "Using Node Fir-10, Vela-7B produced 0.625 latency efficiency score on LumenFacts-3 for Project Saffron on 2026-06-13."}, {"sent_id": 7, "text": "Project Sonata recorded a model choice for error analysis on 2026-05-31."}, {"sent_id": 8, "text": "The chosen model was Marble-7B."}, {"sent_id": 9, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-06-21."}, {"sent_id": 10, "text": "The chosen model was Atlas-8B."}, {"sent_id": 11, "text": "Project Sonata ran Marble-3B on NereidNotes using Node Maple-01 on 2026-06-13."}, {"sent_id": 12, "text": "The reported macro F1 was 0.586."}, {"sent_id": 13, "text": "The staffing note for Project Sonata was dated 2026-06-23."}, {"sent_id": 14, "text": "It named Dr. Jonas Rios for the lead duties."}, {"sent_id": 15, "text": "Cedar-3B was the model selected by Project Meridian on 2026-06-24 for the error analysis track."}, {"sent_id": 16, "text": "Project Anchor received 2 GPUs from Node Cedar-02 in the 2026-04-20 allocation log."}], "source": "synthetic_rewritten", "title": "Run Brief 289"}
{"doc_id": "doc_000290", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For evidence retrieval, the 2026-06-10 note says Project Anchor went with River-8B."}, {"sent_id": 3, "text": "Project Meridian recorded a model choice for error analysis on 2026-05-31."}, {"sent_id": 4, "text": "The chosen model was Cedar-8B."}, {"sent_id": 5, "text": "The 2026-05-23 compute note links Node Cedar-02 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 6, "text": "On 2026-06-03, Project Nereid retired document chunking in favor of metric smoothing."}, {"sent_id": 7, "text": "In the 2026-04-06 staffing section, Project Aster listed Dr. Noah Chen as the retrieval owner."}, {"sent_id": 8, "text": "Project Aster's Milestone T1 deadline change on 2026-04-22 replaced 2026-05-07 with 2026-05-03."}, {"sent_id": 9, "text": "The 2026-06-13 failure note says Orchid-7B on Node Cedar-02 failed for Project Aster because of an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Lab Digest 290"}
{"doc_id": "doc_000291", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor's Milestone H1 deadline change on 2026-04-29 replaced 2026-05-23 with 2026-05-27."}, {"sent_id": 3, "text": "For evidence retrieval, the 2026-05-20 note says Project Saffron went with Aster-7B."}, {"sent_id": 4, "text": "The 2026-05-02 result for Project Meridian put Nimbus-7B on MemoTrace at 0.783 latency efficiency score, with Node Elm-08 handling the run."}, {"sent_id": 5, "text": "Project Aster ran River-7B on RiverBench using Node Rowan-09 on 2026-06-09."}, {"sent_id": 6, "text": "The reported macro F1 was 0.554."}, {"sent_id": 7, "text": "The 2026-05-14 compute note links Node Laurel-11 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 8, "text": "Project Nereid received 3 GPUs from Node Laurel-11 in the 2026-06-08 allocation log."}, {"sent_id": 9, "text": "On 2026-06-03, Project Meridian shifted Milestone P1: 2026-06-21 became 2026-06-27."}, {"sent_id": 10, "text": "Project Saffron received 2 GPUs from Node Fir-10 in the 2026-06-05 allocation log."}, {"sent_id": 11, "text": "On 2026-04-20, Project Sonata shifted Milestone X1: 2026-04-27 became 2026-04-26."}], "source": "synthetic_rewritten", "title": "Project Update 291"}
{"doc_id": "doc_000292", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-25 compute note links Node Spruce-03 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 3, "text": "The 2026-06-04 compute note links Node Hazel-14 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 4, "text": "A compute allocation for Project Aster was logged on 2026-04-11."}, {"sent_id": 5, "text": "Node Cedar-02 supplied 3 GPUs."}, {"sent_id": 6, "text": "In the 2026-05-18 staffing section, Project Sonata listed Rohan Singh as the retrieval owner."}, {"sent_id": 7, "text": "Using Node Spruce-03, Lumen-3B produced 0.518 macro F1 on TraceEval for Project Aster on 2026-04-07."}, {"sent_id": 8, "text": "A compute allocation for Project Meridian was logged on 2026-06-08."}, {"sent_id": 9, "text": "Node Birch-04 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 292"}
{"doc_id": "doc_000293", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-04-01 compute note links Node Hazel-14 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "Using Node Laurel-11, Cedar-8B produced 0.711 accuracy on VestaLogs-2 for Project Meridian on 2026-05-19."}, {"sent_id": 4, "text": "The 2026-06-17 method entry for Project Sonata moved from dense retrieval to rank fusion."}, {"sent_id": 5, "text": "The 2026-06-06 method entry for Project Saffron moved from evidence pooling to calibrated voting."}, {"sent_id": 6, "text": "On 2026-04-29, Project Anchor retired QLoRA adaptation in favor of hybrid retrieval."}, {"sent_id": 7, "text": "The 2026-05-26 result for Project Sonata put Marble-3B on NereidNotes at 0.676 macro F1, with Node Maple-01 handling the run."}, {"sent_id": 8, "text": "On 2026-05-27, Project Nereid shifted Milestone V1: 2026-06-17 became 2026-06-26."}, {"sent_id": 9, "text": "Project Aster recorded a model choice for error analysis on 2026-04-10."}, {"sent_id": 10, "text": "The chosen model was Finch-8B."}], "source": "synthetic_rewritten", "title": "Project Update 293"}
{"doc_id": "doc_000294", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-04-25 compute note links Node Sycamore-13 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "A compute allocation for Project Anchor was logged on 2026-04-01."}, {"sent_id": 4, "text": "Node Spruce-03 supplied 2 GPUs."}, {"sent_id": 5, "text": "Project Meridian ran Atlas-7B on SignalSet-3 using Node Hazel-14 on 2026-06-28."}, {"sent_id": 6, "text": "The reported accuracy was 0.849."}, {"sent_id": 7, "text": "Project Saffron recorded a model choice for error analysis on 2026-06-28."}, {"sent_id": 8, "text": "The chosen model was Marble-8B."}, {"sent_id": 9, "text": "Project Sonata recorded a model choice for claim classification on 2026-05-21."}, {"sent_id": 10, "text": "The chosen model was Quartz-8B."}, {"sent_id": 11, "text": "The 2026-04-30 compute note links Node Spruce-03 to Project Aster with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Lab Digest 294"}
{"doc_id": "doc_000295", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-06-29 result for Project Aster put Lumen-8B on TraceEval at 0.818 macro F1, with Node Spruce-03 handling the run."}, {"sent_id": 3, "text": "Project Saffron logged a personnel update on 2026-04-28: Priya Vale took the evaluation owner responsibilities."}, {"sent_id": 4, "text": "On 2026-05-06, Project Sonata retired data mixing in favor of contrastive tuning."}, {"sent_id": 5, "text": "The review note for Amber Ranking Study 2 was dated 2026-06-10."}, {"sent_id": 6, "text": "It described the method as teacher distillation and said it did not use a reward model."}, {"sent_id": 7, "text": "For Project Aster, the run using Orchid-8B on Node Cedar-02 did not complete on 2026-04-30; the issue was an out-of-memory error."}, {"sent_id": 8, "text": "A compute allocation for Project Meridian was logged on 2026-05-22."}, {"sent_id": 9, "text": "Node Aspen-01 supplied 4 GPUs."}, {"sent_id": 10, "text": "Nova-8B was the model selected by Project Nereid on 2026-04-29 for the reranking track."}, {"sent_id": 11, "text": "Project Nereid had a failed run on 2026-05-22 involving Nova-3B and Node Aspen-01."}, {"sent_id": 12, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 13, "text": "Project Nereid's Milestone P1 deadline change on 2026-05-08 replaced 2026-05-31 with 2026-05-29."}], "source": "synthetic_rewritten", "title": "Project Update 295"}
{"doc_id": "doc_000296", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Anchor received 4 GPUs from Node Poplar-12 in the 2026-06-13 allocation log."}, {"sent_id": 3, "text": "The 2026-05-13 method entry for Project Anchor moved from calibrated voting to confidence calibration."}, {"sent_id": 4, "text": "Milestone T1 for Project Anchor was revised on 2026-06-25."}, {"sent_id": 5, "text": "The schedule moved from 2026-07-05 to 2026-07-15."}, {"sent_id": 6, "text": "Project Nereid ran Nova-7B on LabQA using Node Aspen-01 on 2026-05-18."}, {"sent_id": 7, "text": "The reported evidence F1 was 0.690."}, {"sent_id": 8, "text": "The 2026-04-14 result for Project Sonata put Quartz-3B on OrionBench at 0.652 macro F1, with Node Pine-07 handling the run."}, {"sent_id": 9, "text": "The staffing note for Project Nereid was dated 2026-06-09."}, {"sent_id": 10, "text": "It named Evan Iyer for the retrieval owner duties."}, {"sent_id": 11, "text": "The 2026-05-27 method entry for Project Saffron moved from alias expansion to dense retrieval."}], "source": "synthetic_rewritten", "title": "Lab Digest 296"}
{"doc_id": "doc_000297", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The staffing note for Project Anchor was dated 2026-05-19."}, {"sent_id": 3, "text": "It named Iris Lane for the evaluation owner duties."}, {"sent_id": 4, "text": "Project Sonata logged a personnel update on 2026-05-05: Samir Kwan took the retrieval owner responsibilities."}, {"sent_id": 5, "text": "Project Sonata ran Mira-3B on TraceEval-3 using Node Sycamore-13 on 2026-04-18."}, {"sent_id": 6, "text": "The reported macro F1 was 0.658."}, {"sent_id": 7, "text": "On 2026-04-09, Project Meridian retired cross-encoder reranking in favor of late interaction."}, {"sent_id": 8, "text": "Milestone N1 for Project Anchor was revised on 2026-05-01."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-23 to 2026-05-29."}, {"sent_id": 10, "text": "In the 2026-06-28 staffing section, Project Meridian listed Nadia Singh as the data steward."}, {"sent_id": 11, "text": "Project Saffron had a method update on 2026-06-17."}, {"sent_id": 12, "text": "The old approach was reward reranking, and the replacement was data mixing."}, {"sent_id": 13, "text": "Atlas-8B was the model selected by Project Meridian on 2026-06-11 for the evidence retrieval track."}], "source": "synthetic_rewritten", "title": "Project Update 297"}
{"doc_id": "doc_000298", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-15 failure note says Marble-8B on Node Maple-01 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 3, "text": "Project Anchor recorded a model choice for claim classification on 2026-05-05."}, {"sent_id": 4, "text": "The chosen model was Helix-7B."}, {"sent_id": 5, "text": "The staffing note for Project Anchor was dated 2026-06-14."}, {"sent_id": 6, "text": "It named Kira Frost for the data steward duties."}, {"sent_id": 7, "text": "Project Aster received 1 GPU from Node Spruce-03 in the 2026-06-13 allocation log."}, {"sent_id": 8, "text": "Project Saffron logged a personnel update on 2026-04-20: Petra Adler took the data steward responsibilities."}, {"sent_id": 9, "text": "Project Sonata's Milestone R1 deadline change on 2026-06-10 replaced 2026-06-29 with 2026-06-26."}, {"sent_id": 10, "text": "Milestone F1 for Project Sonata was revised on 2026-06-19."}, {"sent_id": 11, "text": "The schedule moved from 2026-07-02 to 2026-07-11."}, {"sent_id": 12, "text": "In the 2026-06-21 staffing section, Project Meridian listed Nadia Chen as the evaluation owner."}, {"sent_id": 13, "text": "Project Sonata logged a personnel update on 2026-04-07: Rohan Singh took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Lab Digest 298"}
{"doc_id": "doc_000299", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Cedar-3B was the model selected by Project Meridian on 2026-04-02 for the error analysis track."}, {"sent_id": 3, "text": "Orchid-3B was the model selected by Project Aster on 2026-05-24 for the reranking track."}, {"sent_id": 4, "text": "The staffing note for Project Nereid was dated 2026-04-14."}, {"sent_id": 5, "text": "It named Vera Kim for the lead duties."}, {"sent_id": 6, "text": "The 2026-05-16 compute note links Node Willow-05 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 7, "text": "Orchid-8B was the model selected by Project Aster on 2026-05-31 for the reranking track."}, {"sent_id": 8, "text": "The 2026-04-16 failure note says Nova-7B on Node Aspen-01 failed for Project Meridian because of a checkpoint-mismatch error."}, {"sent_id": 9, "text": "Marble-7B was the model selected by Project Saffron on 2026-05-08 for the error analysis track."}, {"sent_id": 10, "text": "For calibration, the 2026-06-21 note says Project Anchor went with Lumen-3B."}], "source": "synthetic_rewritten", "title": "Project Update 299"}
{"doc_id": "doc_000300", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid had a method update on 2026-05-17."}, {"sent_id": 3, "text": "The old approach was query rewriting, and the replacement was sentence pruning."}, {"sent_id": 4, "text": "The 2026-06-13 compute note links Node Rowan-09 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 5, "text": "The 2026-06-27 method entry for Project Sonata moved from rank fusion to teacher distillation."}, {"sent_id": 6, "text": "Using Node Birch-04, Kestrel-8B produced 0.837 accuracy on CedarQA for Project Meridian on 2026-06-07."}, {"sent_id": 7, "text": "Project Meridian received 4 GPUs from Node Aspen-01 in the 2026-04-30 allocation log."}, {"sent_id": 8, "text": "The 2026-04-11 compute note links Node Pine-07 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 9, "text": "Quartz-8B was the model selected by Project Sonata on 2026-06-24 for the claim classification track."}, {"sent_id": 10, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-05-03."}, {"sent_id": 11, "text": "The chosen model was Atlas-7B."}], "source": "synthetic_rewritten", "title": "Review Packet 300"}
{"doc_id": "doc_000301", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Anchor had a failed run on 2026-06-04 involving River-8B and Node Rowan-09."}, {"sent_id": 3, "text": "The recorded cause was a missing-index error."}, {"sent_id": 4, "text": "The 2026-05-19 result for Project Saffron put Quartz-3B on CedarQA-2 at 0.607 latency efficiency score, with Node Pine-07 handling the run."}, {"sent_id": 5, "text": "Project Saffron logged a personnel update on 2026-05-26: Owen Marin took the data steward responsibilities."}, {"sent_id": 6, "text": "Nimbus-7B was the model selected by Project Meridian on 2026-06-14 for the calibration track."}, {"sent_id": 7, "text": "Project Sonata had a method update on 2026-06-06."}, {"sent_id": 8, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 9, "text": "The 2026-04-03 failure note says Cedar-7B on Node Laurel-11 failed for Project Nereid because of an out-of-memory error."}, {"sent_id": 10, "text": "On 2026-06-16, Project Saffron shifted Milestone F1: 2026-06-24 became 2026-07-02."}], "source": "synthetic_rewritten", "title": "Run Brief 301"}
{"doc_id": "doc_000302", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid recorded a model choice for reranking on 2026-06-30."}, {"sent_id": 3, "text": "The chosen model was Nova-7B."}, {"sent_id": 4, "text": "The 2026-05-26 result for Project Sonata put Mira-8B on TraceEval-3 at 0.598 evidence F1, with Node Sycamore-13 handling the run."}, {"sent_id": 5, "text": "For calibration, the 2026-04-05 note says Project Aster went with Lumen-8B."}, {"sent_id": 6, "text": "The 2026-04-12 method entry for Project Aster moved from data mixing to contrastive tuning."}, {"sent_id": 7, "text": "Using Node Rowan-09, River-7B produced 0.899 accuracy on SignalSet-2 for Project Anchor on 2026-05-23."}, {"sent_id": 8, "text": "Vela-3B was the model selected by Project Sonata on 2026-04-16 for the reranking track."}], "source": "synthetic_rewritten", "title": "Review Packet 302"}
{"doc_id": "doc_000303", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-07 result for Project Sonata put Vela-3B on LabQA-2 at 0.700 macro F1, with Node Fir-10 handling the run."}, {"sent_id": 3, "text": "Atlas-7B was the model selected by Project Meridian on 2026-05-10 for the evidence retrieval track."}, {"sent_id": 4, "text": "Project Sonata logged a personnel update on 2026-06-29: Dr. Jonas Rios took the lead responsibilities."}, {"sent_id": 5, "text": "The staffing note for Project Meridian was dated 2026-06-08."}, {"sent_id": 6, "text": "It named Dr. Ravi Nadir for the data steward duties."}, {"sent_id": 7, "text": "Project Nereid logged a personnel update on 2026-06-01: Dr. Julian Stone took the retrieval owner responsibilities."}, {"sent_id": 8, "text": "Project Aster had a method update on 2026-05-30."}, {"sent_id": 9, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}, {"sent_id": 10, "text": "Project Anchor had a method update on 2026-05-23."}, {"sent_id": 11, "text": "The old approach was contrastive tuning, and the replacement was threshold search."}, {"sent_id": 12, "text": "The review note for Amber Ranking Study 1 was dated 2026-06-15."}, {"sent_id": 13, "text": "It described the method as teacher distillation and said it did not use a reward model."}, {"sent_id": 14, "text": "Project Meridian had a method update on 2026-05-02."}, {"sent_id": 15, "text": "The old approach was sentence pruning, and the replacement was cross-encoder reranking."}], "source": "synthetic_rewritten", "title": "Run Brief 303"}
{"doc_id": "doc_000304", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone F1 for Project Sonata was revised on 2026-05-06."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-29 to 2026-05-27."}, {"sent_id": 4, "text": "Project Meridian recorded a model choice for calibration on 2026-05-13."}, {"sent_id": 5, "text": "The chosen model was Nimbus-3B."}, {"sent_id": 6, "text": "Project Aster recorded a model choice for calibration on 2026-05-01."}, {"sent_id": 7, "text": "The chosen model was Lumen-3B."}, {"sent_id": 8, "text": "For error analysis, the 2026-05-24 note says Project Meridian went with Cedar-3B."}, {"sent_id": 9, "text": "Milestone J1 for Project Meridian was revised on 2026-04-30."}, {"sent_id": 10, "text": "The schedule moved from 2026-05-10 to 2026-05-16."}, {"sent_id": 11, "text": "Project Aster recorded a model choice for calibration on 2026-04-12."}, {"sent_id": 12, "text": "The chosen model was Lumen-7B."}, {"sent_id": 13, "text": "The 2026-06-06 method entry for Project Aster moved from alias expansion to dense retrieval."}, {"sent_id": 14, "text": "Project Sonata received 3 GPUs from Node Fir-10 in the 2026-06-20 allocation log."}], "source": "synthetic_rewritten", "title": "Review Packet 304"}
{"doc_id": "doc_000305", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The staffing note for Project Sonata was dated 2026-05-19."}, {"sent_id": 3, "text": "It named Rohan Park for the lead duties."}, {"sent_id": 4, "text": "Quartz-3B was the model selected by Project Sonata on 2026-05-06 for the claim classification track."}, {"sent_id": 5, "text": "Project Aster's Milestone T1 deadline change on 2026-05-14 replaced 2026-06-08 with 2026-06-17."}, {"sent_id": 6, "text": "Project Anchor received 2 GPUs from Node Poplar-12 in the 2026-05-11 allocation log."}, {"sent_id": 7, "text": "The 2026-06-01 compute note links Node Sycamore-13 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 8, "text": "Project Nereid recorded a model choice for calibration on 2026-05-03."}, {"sent_id": 9, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 10, "text": "For Project Anchor, the run using Orchid-8B on Node Cedar-02 did not complete on 2026-05-29; the issue was a missing-index error."}, {"sent_id": 11, "text": "The 2026-05-23 compute note links Node Fir-10 to Project Sonata with an allocation of 3 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 305"}
{"doc_id": "doc_000306", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-05-27, Project Anchor retired dense retrieval in favor of rank fusion."}, {"sent_id": 3, "text": "Lumen-3B was the model selected by Project Anchor on 2026-04-15 for the calibration track."}, {"sent_id": 4, "text": "Project Nereid had a failed run on 2026-05-14 involving Nova-8B and Node Aspen-01."}, {"sent_id": 5, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 6, "text": "Project Aster recorded a model choice for error analysis on 2026-04-30."}, {"sent_id": 7, "text": "The chosen model was Finch-3B."}, {"sent_id": 8, "text": "The staffing note for Project Aster was dated 2026-04-06."}, {"sent_id": 9, "text": "It named Talia Reed for the retrieval owner duties."}, {"sent_id": 10, "text": "For reranking, the 2026-06-05 note says Project Aster went with Orchid-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 306"}
{"doc_id": "doc_000307", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Sonata had a method update on 2026-05-16."}, {"sent_id": 3, "text": "The old approach was document chunking, and the replacement was metric smoothing."}, {"sent_id": 4, "text": "The review note for Quiet Retriever Study 1 was dated 2026-05-29."}, {"sent_id": 5, "text": "It described the method as structured prompting and said it did not use a reward model."}, {"sent_id": 6, "text": "Project Aster recorded a model choice for calibration on 2026-04-24."}, {"sent_id": 7, "text": "The chosen model was Lumen-7B."}, {"sent_id": 8, "text": "For claim classification, the 2026-06-12 note says Project Aster went with Helix-3B."}, {"sent_id": 9, "text": "For claim classification, the 2026-05-03 note says Project Meridian went with Kestrel-3B."}, {"sent_id": 10, "text": "Project Nereid received 1 GPU from Node Laurel-11 in the 2026-05-08 allocation log."}, {"sent_id": 11, "text": "Project Sonata logged a personnel update on 2026-05-11: Samir Kwan took the retrieval owner responsibilities."}, {"sent_id": 12, "text": "Project Meridian had a method update on 2026-05-03."}, {"sent_id": 13, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 14, "text": "The staffing note for Project Saffron was dated 2026-06-02."}, {"sent_id": 15, "text": "It named Owen Torres for the evaluation owner duties."}], "source": "synthetic_rewritten", "title": "Run Brief 307"}
{"doc_id": "doc_000308", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-05-17 result for Project Saffron put Marble-8B on VestaLogs-3 at 0.721 latency efficiency score, with Node Maple-01 handling the run."}, {"sent_id": 3, "text": "Project Meridian recorded a model choice for reranking on 2026-05-21."}, {"sent_id": 4, "text": "The chosen model was Nova-8B."}, {"sent_id": 5, "text": "Project Meridian had a failed run on 2026-04-18 involving Cedar-3B and Node Laurel-11."}, {"sent_id": 6, "text": "The recorded cause was a missing-index error."}, {"sent_id": 7, "text": "The 2026-04-25 compute note links Node Aspen-01 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 8, "text": "Project Meridian had a failed run on 2026-05-30 involving Nova-3B and Node Aspen-01."}, {"sent_id": 9, "text": "The recorded cause was a missing-index error."}, {"sent_id": 10, "text": "Project Nereid ran Cedar-3B on NereidNotes-3 using Node Laurel-11 on 2026-06-07."}, {"sent_id": 11, "text": "The reported macro F1 was 0.876."}, {"sent_id": 12, "text": "For Project Anchor, the run using Orchid-3B on Node Cedar-02 did not complete on 2026-05-28; the issue was a checkpoint-mismatch error."}, {"sent_id": 13, "text": "Using Node Willow-05, Aster-7B produced 0.859 accuracy on SignalSet for Project Saffron on 2026-05-05."}], "source": "synthetic_rewritten", "title": "Review Packet 308"}
{"doc_id": "doc_000309", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-26, Project Sonata retired hard-negative mining in favor of LoRA adaptation."}, {"sent_id": 3, "text": "On 2026-05-15, Project Saffron shifted Milestone R1: 2026-06-08 became 2026-06-16."}, {"sent_id": 4, "text": "Milestone B2 for Project Meridian was revised on 2026-06-08."}, {"sent_id": 5, "text": "The schedule moved from 2026-07-02 to 2026-07-06."}, {"sent_id": 6, "text": "Project Meridian recorded a model choice for reranking on 2026-05-08."}, {"sent_id": 7, "text": "The chosen model was Nova-3B."}, {"sent_id": 8, "text": "For error analysis, the 2026-04-05 note says Project Meridian went with Cedar-7B."}, {"sent_id": 9, "text": "Project Meridian recorded a model choice for calibration on 2026-04-09."}, {"sent_id": 10, "text": "The chosen model was Nimbus-3B."}, {"sent_id": 11, "text": "On 2026-06-12, the Amber Ranking Study 4 review reported late interaction and said it did not use a reward model."}, {"sent_id": 12, "text": "Finch-8B was the model selected by Project Anchor on 2026-06-05 for the error analysis track."}], "source": "synthetic_rewritten", "title": "Run Brief 309"}
{"doc_id": "doc_000310", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-06-04 compute note links Node Poplar-12 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 3, "text": "Project Aster ran Orchid-8B on LabQA-3 using Node Cedar-02 on 2026-05-09."}, {"sent_id": 4, "text": "The reported evidence F1 was 0.800."}, {"sent_id": 5, "text": "The 2026-06-14 method entry for Project Saffron moved from teacher distillation to query rewriting."}, {"sent_id": 6, "text": "Project Aster received 3 GPUs from Node Rowan-09 in the 2026-06-25 allocation log."}, {"sent_id": 7, "text": "Aster-7B was the model selected by Project Saffron on 2026-04-10 for the evidence retrieval track."}, {"sent_id": 8, "text": "The 2026-05-30 method entry for Project Saffron moved from teacher distillation to query rewriting."}, {"sent_id": 9, "text": "For claim classification, the 2026-04-14 note says Project Nereid went with Kestrel-3B."}, {"sent_id": 10, "text": "On 2026-05-06, Project Nereid shifted Milestone J1: 2026-05-31 became 2026-06-05."}], "source": "synthetic_rewritten", "title": "Lab Digest 310"}
{"doc_id": "doc_000311", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "A compute allocation for Project Anchor was logged on 2026-04-29."}, {"sent_id": 3, "text": "Node Cedar-02 supplied 4 GPUs."}, {"sent_id": 4, "text": "The 2026-05-28 failure note says Atlas-7B on Node Hazel-14 failed for Project Meridian because of a checkpoint-mismatch error."}, {"sent_id": 5, "text": "Project Anchor logged a personnel update on 2026-04-14: Iris Lane took the evaluation owner responsibilities."}, {"sent_id": 6, "text": "Project Sonata had a method update on 2026-06-21."}, {"sent_id": 7, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 8, "text": "The 2026-05-08 failure note says Finch-3B on Node Juniper-06 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 9, "text": "The 2026-04-17 compute note links Node Laurel-11 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 10, "text": "On 2026-06-08, Project Anchor shifted Milestone B1: 2026-07-04 became 2026-07-14."}, {"sent_id": 11, "text": "In the 2026-05-31 staffing section, Project Nereid listed Vera Kim as the lead."}, {"sent_id": 12, "text": "Project Sonata had a failed run on 2026-06-18 involving Vela-7B and Node Fir-10."}, {"sent_id": 13, "text": "The recorded cause was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Project Update 311"}
{"doc_id": "doc_000312", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The staffing note for Project Saffron was dated 2026-06-01."}, {"sent_id": 3, "text": "It named Lena Sol for the evaluation owner duties."}, {"sent_id": 4, "text": "In the 2026-06-22 staffing section, Project Aster listed Talia Marin as the lead."}, {"sent_id": 5, "text": "The 2026-06-20 result for Project Saffron put Quartz-7B on CedarQA-2 at 0.577 latency efficiency score, with Node Pine-07 handling the run."}, {"sent_id": 6, "text": "The 2026-04-02 failure note says Marble-7B on Node Maple-01 failed for Project Saffron because of a missing-index error."}, {"sent_id": 7, "text": "The staffing note for Project Aster was dated 2026-05-26."}, {"sent_id": 8, "text": "It named Talia Reed for the retrieval owner duties."}, {"sent_id": 9, "text": "On 2026-04-02, Project Nereid retired evidence pooling in favor of calibrated voting."}, {"sent_id": 10, "text": "Project Meridian received 4 GPUs from Node Elm-08 in the 2026-04-17 allocation log."}, {"sent_id": 11, "text": "Project Meridian's Milestone J1 deadline change on 2026-04-01 replaced 2026-04-11 with 2026-04-21."}, {"sent_id": 12, "text": "Project Saffron's Milestone X1 deadline change on 2026-04-17 replaced 2026-04-29 with 2026-05-03."}], "source": "synthetic_rewritten", "title": "Lab Digest 312"}
{"doc_id": "doc_000313", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid had a failed run on 2026-05-29 involving Nova-8B and Node Aspen-01."}, {"sent_id": 3, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 4, "text": "Project Anchor recorded a model choice for claim classification on 2026-06-10."}, {"sent_id": 5, "text": "The chosen model was Helix-7B."}, {"sent_id": 6, "text": "Pale Compass Study 2 appeared in a 2026-04-06 review entry using temporal filtering; the note said it did not use a reward model."}, {"sent_id": 7, "text": "A compute allocation for Project Aster was logged on 2026-04-18."}, {"sent_id": 8, "text": "Node Cedar-02 supplied 1 GPU."}, {"sent_id": 9, "text": "The 2026-05-28 compute note links Node Fir-10 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 10, "text": "Project Anchor logged a personnel update on 2026-04-20: Leo Hale took the data steward responsibilities."}, {"sent_id": 11, "text": "Milestone L1 for Project Sonata was revised on 2026-05-15."}, {"sent_id": 12, "text": "The schedule moved from 2026-05-26 to 2026-05-21."}, {"sent_id": 13, "text": "Nova-8B was the model selected by Project Meridian on 2026-06-03 for the reranking track."}], "source": "synthetic_rewritten", "title": "Project Update 313"}
{"doc_id": "doc_000314", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-06-02 result for Project Aster put Finch-7B on NereidNotes-2 at 0.836 macro F1, with Node Juniper-06 handling the run."}, {"sent_id": 3, "text": "A compute allocation for Project Anchor was logged on 2026-05-22."}, {"sent_id": 4, "text": "Node Juniper-06 supplied 4 GPUs."}, {"sent_id": 5, "text": "Project Nereid logged a personnel update on 2026-06-09: Dr. Julian Stone took the retrieval owner responsibilities."}, {"sent_id": 6, "text": "A compute allocation for Project Sonata was logged on 2026-04-25."}, {"sent_id": 7, "text": "Node Willow-05 supplied 1 GPU."}, {"sent_id": 8, "text": "A compute allocation for Project Aster was logged on 2026-05-14."}, {"sent_id": 9, "text": "Node Cedar-02 supplied 3 GPUs."}, {"sent_id": 10, "text": "Milestone Z1 for Project Anchor was revised on 2026-04-20."}, {"sent_id": 11, "text": "The schedule moved from 2026-05-08 to 2026-05-18."}, {"sent_id": 12, "text": "For Project Saffron, the run using Vela-3B on Node Fir-10 did not complete on 2026-05-15; the issue was a checkpoint-mismatch error."}, {"sent_id": 13, "text": "The staffing note for Project Nereid was dated 2026-05-18."}, {"sent_id": 14, "text": "It named Darian Grant for the lead duties."}, {"sent_id": 15, "text": "Project Sonata had a method update on 2026-04-30."}, {"sent_id": 16, "text": "The old approach was metric smoothing, and the replacement was hard-negative mining."}], "source": "synthetic_rewritten", "title": "Lab Digest 314"}
{"doc_id": "doc_000315", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata's Milestone D2 deadline change on 2026-05-01 replaced 2026-05-26 with 2026-06-04."}, {"sent_id": 3, "text": "The 2026-06-24 compute note links Node Poplar-12 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 4, "text": "On 2026-04-01, Project Sonata shifted Milestone L1: 2026-04-22 became 2026-04-27."}, {"sent_id": 5, "text": "On 2026-04-08, Project Aster shifted Milestone T1: 2026-04-23 became 2026-04-19."}, {"sent_id": 6, "text": "Project Meridian's Milestone B2 deadline change on 2026-06-10 replaced 2026-06-24 with 2026-06-30."}, {"sent_id": 7, "text": "For evidence retrieval, the 2026-05-21 note says Project Sonata went with Aster-7B."}, {"sent_id": 8, "text": "For Project Meridian, the run using Cedar-8B on Node Laurel-11 did not complete on 2026-06-12; the issue was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "Cedar-7B was the model selected by Project Nereid on 2026-04-12 for the error analysis track."}, {"sent_id": 10, "text": "Milestone N1 for Project Aster was revised on 2026-05-13."}, {"sent_id": 11, "text": "The schedule moved from 2026-05-30 to 2026-06-08."}], "source": "synthetic_rewritten", "title": "Project Update 315"}
{"doc_id": "doc_000316", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Atlas-3B was the model selected by Project Meridian on 2026-04-08 for the evidence retrieval track."}, {"sent_id": 3, "text": "A compute allocation for Project Meridian was logged on 2026-04-15."}, {"sent_id": 4, "text": "Node Laurel-11 supplied 2 GPUs."}, {"sent_id": 5, "text": "For calibration, the 2026-04-16 note says Project Meridian went with Nimbus-8B."}, {"sent_id": 6, "text": "In the 2026-05-18 staffing section, Project Nereid listed Nora Sol as the retrieval owner."}, {"sent_id": 7, "text": "The 2026-05-08 failure note says Nimbus-3B on Node Elm-08 failed for Project Nereid because of an out-of-memory error."}, {"sent_id": 8, "text": "On 2026-06-12, Project Meridian shifted Milestone D1: 2026-06-24 became 2026-07-02."}, {"sent_id": 9, "text": "On 2026-06-10, Project Saffron shifted Milestone F1: 2026-06-28 became 2026-07-08."}], "source": "synthetic_rewritten", "title": "Lab Digest 316"}
{"doc_id": "doc_000317", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Aster logged a personnel update on 2026-06-15: Talia Reed took the retrieval owner responsibilities."}, {"sent_id": 3, "text": "Using Node Cedar-02, Orchid-3B produced 0.725 accuracy on LumenFacts for Project Anchor on 2026-04-28."}, {"sent_id": 4, "text": "Using Node Poplar-12, Helix-8B produced 0.527 accuracy on CedarQA-3 for Project Anchor on 2026-06-30."}, {"sent_id": 5, "text": "On 2026-06-11, Project Nereid shifted Milestone B2: 2026-06-20 became 2026-06-25."}, {"sent_id": 6, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-04-16."}, {"sent_id": 7, "text": "The chosen model was Aster-7B."}, {"sent_id": 8, "text": "The 2026-06-18 compute note links Node Pine-07 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 9, "text": "Project Sonata logged a personnel update on 2026-06-08: Mara Lane took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Project Update 317"}
{"doc_id": "doc_000318", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Sonata recorded a model choice for calibration on 2026-04-16."}, {"sent_id": 3, "text": "The chosen model was Mira-8B."}, {"sent_id": 4, "text": "The 2026-06-22 compute note links Node Elm-08 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 5, "text": "A compute allocation for Project Meridian was logged on 2026-05-23."}, {"sent_id": 6, "text": "Node Elm-08 supplied 4 GPUs."}, {"sent_id": 7, "text": "The 2026-06-13 compute note links Node Maple-01 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 8, "text": "Project Meridian received 2 GPUs from Node Hazel-14 in the 2026-06-12 allocation log."}, {"sent_id": 9, "text": "Using Node Poplar-12, Helix-7B produced 0.512 evidence F1 on OrionBench-2 for Project Aster on 2026-05-12."}, {"sent_id": 10, "text": "The staffing note for Project Nereid was dated 2026-04-27."}, {"sent_id": 11, "text": "It named Dr. Julian Stone for the retrieval owner duties."}], "source": "synthetic_rewritten", "title": "Lab Digest 318"}
{"doc_id": "doc_000319", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Anchor was logged on 2026-04-03."}, {"sent_id": 3, "text": "Node Spruce-03 supplied 2 GPUs."}, {"sent_id": 4, "text": "For Project Saffron, the run using Quartz-3B on Node Pine-07 did not complete on 2026-05-22; the issue was a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-05-30 method entry for Project Sonata moved from temporal filtering to alias expansion."}, {"sent_id": 6, "text": "On 2026-04-01, Project Saffron shifted Milestone R1: 2026-04-15 became 2026-04-21."}, {"sent_id": 7, "text": "Project Sonata ran Quartz-8B on OrionBench using Node Pine-07 on 2026-06-30."}, {"sent_id": 8, "text": "The reported macro F1 was 0.592."}, {"sent_id": 9, "text": "Project Sonata received 3 GPUs from Node Maple-01 in the 2026-05-23 allocation log."}, {"sent_id": 10, "text": "Using Node Elm-08, Nimbus-8B produced 0.798 macro F1 on TraceEval-2 for Project Nereid on 2026-05-12."}, {"sent_id": 11, "text": "Project Saffron received 2 GPUs from Node Sycamore-13 in the 2026-05-15 allocation log."}], "source": "synthetic_rewritten", "title": "Project Update 319"}
{"doc_id": "doc_000320", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian recorded a model choice for calibration on 2026-04-26."}, {"sent_id": 3, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 4, "text": "The 2026-05-02 compute note links Node Cedar-02 to Project Aster with an allocation of 1 GPU."}, {"sent_id": 5, "text": "Project Anchor logged a personnel update on 2026-06-08: Iris Lane took the evaluation owner responsibilities."}, {"sent_id": 6, "text": "Project Aster received 3 GPUs from Node Juniper-06 in the 2026-05-06 allocation log."}, {"sent_id": 7, "text": "On 2026-06-09, Project Meridian shifted Milestone B2: 2026-06-23 became 2026-06-29."}, {"sent_id": 8, "text": "On 2026-06-24, Project Saffron shifted Milestone F1: 2026-07-12 became 2026-07-18."}, {"sent_id": 9, "text": "A compute allocation for Project Anchor was logged on 2026-06-26."}, {"sent_id": 10, "text": "Node Cedar-02 supplied 2 GPUs."}, {"sent_id": 11, "text": "Project Saffron had a failed run on 2026-05-21 involving Quartz-7B and Node Pine-07."}, {"sent_id": 12, "text": "The recorded cause was a missing-index error."}, {"sent_id": 13, "text": "For Project Saffron, the run using Aster-3B on Node Willow-05 did not complete on 2026-06-05; the issue was a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 320"}
{"doc_id": "doc_000321", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-05 result for Project Nereid put Nimbus-3B on TraceEval-2 at 0.768 evidence F1, with Node Elm-08 handling the run."}, {"sent_id": 3, "text": "On 2026-04-16, Project Saffron shifted Milestone X1: 2026-05-08 became 2026-05-18."}, {"sent_id": 4, "text": "Project Sonata had a method update on 2026-05-17."}, {"sent_id": 5, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 6, "text": "On 2026-05-07, Project Aster shifted Milestone H1: 2026-05-16 became 2026-05-25."}, {"sent_id": 7, "text": "Project Aster's Milestone B1 deadline change on 2026-04-29 replaced 2026-05-10 with 2026-05-05."}, {"sent_id": 8, "text": "The 2026-06-13 compute note links Node Aspen-01 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 9, "text": "Project Nereid's Milestone B2 deadline change on 2026-04-28 replaced 2026-05-07 with 2026-05-12."}], "source": "synthetic_rewritten", "title": "Review Packet 321"}
{"doc_id": "doc_000322", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor logged a personnel update on 2026-04-27: Kira Iyer took the evaluation owner responsibilities."}, {"sent_id": 3, "text": "A compute allocation for Project Anchor was logged on 2026-04-02."}, {"sent_id": 4, "text": "Node Cedar-02 supplied 2 GPUs."}, {"sent_id": 5, "text": "On 2026-06-17, Project Aster shifted Milestone Z1: 2026-06-30 became 2026-07-09."}, {"sent_id": 6, "text": "Project Anchor recorded a model choice for evidence retrieval on 2026-05-19."}, {"sent_id": 7, "text": "The chosen model was River-3B."}, {"sent_id": 8, "text": "The 2026-05-14 method entry for Project Aster moved from teacher distillation to query rewriting."}, {"sent_id": 9, "text": "Quartz-7B was the model selected by Project Sonata on 2026-05-07 for the claim classification track."}, {"sent_id": 10, "text": "Project Saffron recorded a model choice for reranking on 2026-05-29."}, {"sent_id": 11, "text": "The chosen model was Vela-7B."}], "source": "synthetic_rewritten", "title": "Run Brief 322"}
{"doc_id": "doc_000323", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-22, Project Anchor shifted Milestone Z1: 2026-04-30 became 2026-05-04."}, {"sent_id": 3, "text": "The 2026-06-23 result for Project Sonata put Vela-8B on LabQA-2 at 0.640 macro F1, with Node Fir-10 handling the run."}, {"sent_id": 4, "text": "The 2026-04-08 method entry for Project Nereid moved from QLoRA adaptation to hybrid retrieval."}, {"sent_id": 5, "text": "On 2026-05-25, Project Nereid shifted Milestone V1: 2026-06-05 became 2026-05-31."}, {"sent_id": 6, "text": "On 2026-04-11, Project Meridian retired LoRA adaptation in favor of alias expansion."}, {"sent_id": 7, "text": "Project Anchor had a method update on 2026-05-27."}, {"sent_id": 8, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}], "source": "synthetic_rewritten", "title": "Review Packet 323"}
{"doc_id": "doc_000324", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The staffing note for Project Anchor was dated 2026-06-16."}, {"sent_id": 3, "text": "It named Kira Iyer for the evaluation owner duties."}, {"sent_id": 4, "text": "Milestone T1 for Project Anchor was revised on 2026-04-16."}, {"sent_id": 5, "text": "The schedule moved from 2026-05-06 to 2026-05-10."}, {"sent_id": 6, "text": "Project Sonata ran Aster-8B on RiverBench-3 using Node Willow-05 on 2026-05-11."}, {"sent_id": 7, "text": "The reported macro F1 was 0.634."}, {"sent_id": 8, "text": "In the 2026-05-25 staffing section, Project Anchor listed Dr. Arun Bauer as the evaluation owner."}, {"sent_id": 9, "text": "For Project Saffron, the run using Aster-8B on Node Willow-05 did not complete on 2026-04-25; the issue was a checkpoint-mismatch error."}, {"sent_id": 10, "text": "The staffing note for Project Saffron was dated 2026-04-27."}, {"sent_id": 11, "text": "It named Priya Moss for the data steward duties."}, {"sent_id": 12, "text": "For Project Meridian, the run using Cedar-8B on Node Laurel-11 did not complete on 2026-04-16; the issue was a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 324"}
{"doc_id": "doc_000325", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-06-13, Project Meridian retired threshold search in favor of reward reranking."}, {"sent_id": 3, "text": "The 2026-04-29 method entry for Project Saffron moved from structured prompting to QLoRA adaptation."}, {"sent_id": 4, "text": "The 2026-05-14 compute note links Node Sycamore-13 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 5, "text": "Project Sonata received 3 GPUs from Node Pine-07 in the 2026-06-27 allocation log."}, {"sent_id": 6, "text": "Project Anchor logged a personnel update on 2026-06-09: Dr. Arun Bauer took the evaluation owner responsibilities."}, {"sent_id": 7, "text": "Project Saffron recorded a model choice for claim classification on 2026-06-19."}, {"sent_id": 8, "text": "The chosen model was Quartz-8B."}, {"sent_id": 9, "text": "Milestone P1 for Project Nereid was revised on 2026-06-29."}, {"sent_id": 10, "text": "The schedule moved from 2026-07-12 to 2026-07-21."}], "source": "synthetic_rewritten", "title": "Review Packet 325"}
{"doc_id": "doc_000326", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-06-01 compute note links Node Pine-07 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 3, "text": "Project Nereid ran Atlas-7B on RiverBench-2 using Node Hazel-14 on 2026-06-29."}, {"sent_id": 4, "text": "The reported evidence F1 was 0.714."}, {"sent_id": 5, "text": "Project Nereid ran Kestrel-8B on OrionBench-3 using Node Birch-04 on 2026-04-11."}, {"sent_id": 6, "text": "The reported macro F1 was 0.732."}, {"sent_id": 7, "text": "The 2026-06-16 result for Project Aster put Finch-8B on NereidNotes-2 at 0.896 macro F1, with Node Juniper-06 handling the run."}, {"sent_id": 8, "text": "Using Node Laurel-11, Cedar-8B produced 0.756 evidence F1 on NereidNotes-3 for Project Nereid on 2026-04-14."}, {"sent_id": 9, "text": "The 2026-05-02 compute note links Node Elm-08 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 10, "text": "On 2026-06-28, Project Nereid retired document chunking in favor of metric smoothing."}, {"sent_id": 11, "text": "Marble-3B was the model selected by Project Sonata on 2026-04-29 for the error analysis track."}, {"sent_id": 12, "text": "For error analysis, the 2026-05-13 note says Project Meridian went with Cedar-8B."}], "source": "synthetic_rewritten", "title": "Run Brief 326"}
{"doc_id": "doc_000327", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian had a failed run on 2026-04-10 involving Nova-7B and Node Aspen-01."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "Project Saffron logged a personnel update on 2026-05-12: Lena Sol took the evaluation owner responsibilities."}, {"sent_id": 5, "text": "The 2026-06-26 compute note links Node Fir-10 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 6, "text": "On 2026-06-04, Project Anchor retired QLoRA adaptation in favor of hybrid retrieval."}, {"sent_id": 7, "text": "For reranking, the 2026-05-17 note says Project Meridian went with Nova-7B."}, {"sent_id": 8, "text": "The 2026-06-30 result for Project Saffron put Vela-3B on LumenFacts-3 at 0.865 latency efficiency score, with Node Fir-10 handling the run."}, {"sent_id": 9, "text": "In the 2026-04-14 staffing section, Project Sonata listed Mara Lane as the retrieval owner."}, {"sent_id": 10, "text": "The 2026-05-28 compute note links Node Willow-05 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 11, "text": "For evidence retrieval, the 2026-04-26 note says Project Sonata went with Aster-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 327"}
{"doc_id": "doc_000328", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron ran Marble-3B on VestaLogs-3 using Node Maple-01 on 2026-05-19."}, {"sent_id": 3, "text": "The reported latency efficiency score was 0.841."}, {"sent_id": 4, "text": "Project Anchor received 4 GPUs from Node Spruce-03 in the 2026-06-12 allocation log."}, {"sent_id": 5, "text": "For evidence retrieval, the 2026-04-05 note says Project Aster went with River-3B."}, {"sent_id": 6, "text": "For error analysis, the 2026-06-14 note says Project Meridian went with Cedar-3B."}, {"sent_id": 7, "text": "Project Anchor recorded a model choice for error analysis on 2026-06-07."}, {"sent_id": 8, "text": "The chosen model was Finch-7B."}, {"sent_id": 9, "text": "Project Anchor's Milestone Z1 deadline change on 2026-04-15 replaced 2026-05-03 with 2026-05-09."}], "source": "synthetic_rewritten", "title": "Run Brief 328"}
{"doc_id": "doc_000329", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The staffing note for Project Aster was dated 2026-04-12."}, {"sent_id": 3, "text": "It named Felix Brooks for the retrieval owner duties."}, {"sent_id": 4, "text": "Project Aster recorded a model choice for calibration on 2026-05-14."}, {"sent_id": 5, "text": "The chosen model was Lumen-8B."}, {"sent_id": 6, "text": "Project Saffron ran Aster-3B on SignalSet using Node Willow-05 on 2026-06-08."}, {"sent_id": 7, "text": "The reported accuracy was 0.559."}, {"sent_id": 8, "text": "Milestone B2 for Project Meridian was revised on 2026-06-02."}, {"sent_id": 9, "text": "The schedule moved from 2026-06-26 to 2026-07-04."}, {"sent_id": 10, "text": "Project Nereid received 3 GPUs from Node Birch-04 in the 2026-06-06 allocation log."}, {"sent_id": 11, "text": "In the 2026-05-11 staffing section, Project Saffron listed Lena Costa as the data steward."}, {"sent_id": 12, "text": "A compute allocation for Project Meridian was logged on 2026-04-22."}, {"sent_id": 13, "text": "Node Laurel-11 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Review Packet 329"}
{"doc_id": "doc_000330", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For evidence retrieval, the 2026-05-13 note says Project Aster went with River-7B."}, {"sent_id": 3, "text": "Milestone J1 for Project Nereid was revised on 2026-06-18."}, {"sent_id": 4, "text": "The schedule moved from 2026-07-13 to 2026-07-18."}, {"sent_id": 5, "text": "On 2026-06-17, Project Sonata retired hybrid retrieval in favor of document chunking."}, {"sent_id": 6, "text": "Project Anchor's Milestone N1 deadline change on 2026-04-01 replaced 2026-04-13 with 2026-04-21."}, {"sent_id": 7, "text": "On 2026-05-02, Project Nereid retired contrastive tuning in favor of threshold search."}, {"sent_id": 8, "text": "For Project Meridian, the run using Nova-8B on Node Aspen-01 did not complete on 2026-04-09; the issue was a missing-index error."}, {"sent_id": 9, "text": "The staffing note for Project Nereid was dated 2026-05-12."}, {"sent_id": 10, "text": "It named Dr. Julian Gray for the lead duties."}, {"sent_id": 11, "text": "Project Meridian ran Kestrel-8B on CedarQA using Node Birch-04 on 2026-06-06."}, {"sent_id": 12, "text": "The reported accuracy was 0.777."}], "source": "synthetic_rewritten", "title": "Project Update 330"}
{"doc_id": "doc_000331", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian received 2 GPUs from Node Birch-04 in the 2026-06-17 allocation log."}, {"sent_id": 3, "text": "On 2026-06-26, the Quiet Retriever Study 2 review reported dense retrieval and said it did not use a reward model."}, {"sent_id": 4, "text": "The staffing note for Project Nereid was dated 2026-05-05."}, {"sent_id": 5, "text": "It named Dr. Julian Stone for the retrieval owner duties."}, {"sent_id": 6, "text": "Project Aster had a failed run on 2026-06-11 involving River-8B and Node Rowan-09."}, {"sent_id": 7, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 8, "text": "The 2026-05-09 result for Project Anchor put River-3B on SignalSet-2 at 0.839 accuracy, with Node Rowan-09 handling the run."}, {"sent_id": 9, "text": "Using Node Birch-04, Kestrel-8B produced 0.657 accuracy on CedarQA for Project Meridian on 2026-04-13."}], "source": "synthetic_rewritten", "title": "Lab Digest 331"}
{"doc_id": "doc_000332", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid ran Nova-8B on LabQA using Node Aspen-01 on 2026-05-26."}, {"sent_id": 3, "text": "The reported evidence F1 was 0.780."}, {"sent_id": 4, "text": "A compute allocation for Project Sonata was logged on 2026-05-08."}, {"sent_id": 5, "text": "Node Pine-07 supplied 3 GPUs."}, {"sent_id": 6, "text": "The review note for Quiet Retriever Study 1 was dated 2026-04-01."}, {"sent_id": 7, "text": "It described the method as dense retrieval and said it did not use a reward model."}, {"sent_id": 8, "text": "Project Nereid had a method update on 2026-06-24."}, {"sent_id": 9, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 10, "text": "The staffing note for Project Anchor was dated 2026-04-21."}, {"sent_id": 11, "text": "It named Kira Iyer for the evaluation owner duties."}, {"sent_id": 12, "text": "Project Nereid recorded a model choice for calibration on 2026-04-08."}, {"sent_id": 13, "text": "The chosen model was Nimbus-3B."}], "source": "synthetic_rewritten", "title": "Project Update 332"}
{"doc_id": "doc_000333", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-04 result for Project Nereid put Nimbus-7B on TraceEval-2 at 0.708 macro F1, with Node Elm-08 handling the run."}, {"sent_id": 3, "text": "Helix-8B was the model selected by Project Anchor on 2026-06-21 for the claim classification track."}, {"sent_id": 4, "text": "Using Node Spruce-03, Lumen-8B produced 0.893 accuracy on MemoTrace-3 for Project Anchor on 2026-05-19."}, {"sent_id": 5, "text": "Project Sonata received 3 GPUs from Node Sycamore-13 in the 2026-05-30 allocation log."}, {"sent_id": 6, "text": "The 2026-06-14 method entry for Project Nereid moved from hard-negative mining to LoRA adaptation."}, {"sent_id": 7, "text": "Project Aster had a method update on 2026-05-10."}, {"sent_id": 8, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}], "source": "synthetic_rewritten", "title": "Lab Digest 333"}
{"doc_id": "doc_000334", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-06-08 staffing section, Project Anchor listed Selene Rios as the data steward."}, {"sent_id": 3, "text": "Project Anchor recorded a model choice for calibration on 2026-04-29."}, {"sent_id": 4, "text": "The chosen model was Lumen-7B."}, {"sent_id": 5, "text": "On 2026-05-13, Project Meridian shifted Milestone D1: 2026-06-04 became 2026-06-14."}, {"sent_id": 6, "text": "River-7B was the model selected by Project Aster on 2026-04-09 for the evidence retrieval track."}, {"sent_id": 7, "text": "Helix-7B was the model selected by Project Aster on 2026-04-15 for the claim classification track."}, {"sent_id": 8, "text": "On 2026-04-17, Project Aster shifted Milestone Z1: 2026-05-10 became 2026-05-08."}, {"sent_id": 9, "text": "Milestone J1 for Project Nereid was revised on 2026-05-13."}, {"sent_id": 10, "text": "The schedule moved from 2026-05-28 to 2026-05-24."}], "source": "synthetic_rewritten", "title": "Project Update 334"}
{"doc_id": "doc_000335", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-04-25 compute note links Node Pine-07 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 3, "text": "Atlas-8B was the model selected by Project Nereid on 2026-06-28 for the evidence retrieval track."}, {"sent_id": 4, "text": "Project Anchor ran Finch-3B on VestaLogs using Node Juniper-06 on 2026-06-14."}, {"sent_id": 5, "text": "The reported accuracy was 0.581."}, {"sent_id": 6, "text": "Milestone P1 for Project Meridian was revised on 2026-04-10."}, {"sent_id": 7, "text": "The schedule moved from 2026-04-28 to 2026-05-08."}, {"sent_id": 8, "text": "The 2026-06-08 result for Project Sonata put Mira-7B on TraceEval-3 at 0.598 macro F1, with Node Sycamore-13 handling the run."}, {"sent_id": 9, "text": "The 2026-05-16 compute note links Node Cedar-02 to Project Anchor with an allocation of 2 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 335"}
{"doc_id": "doc_000336", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For Project Aster, the run using Helix-8B on Node Poplar-12 did not complete on 2026-05-14; the issue was an unstable-validation-loss error."}, {"sent_id": 3, "text": "Milestone H1 for Project Anchor was revised on 2026-05-11."}, {"sent_id": 4, "text": "The schedule moved from 2026-05-25 to 2026-05-31."}, {"sent_id": 5, "text": "Project Nereid logged a personnel update on 2026-06-23: Evan Iyer took the retrieval owner responsibilities."}, {"sent_id": 6, "text": "On 2026-06-21, Project Aster retired reward reranking in favor of data mixing."}, {"sent_id": 7, "text": "Project Anchor had a method update on 2026-05-17."}, {"sent_id": 8, "text": "The old approach was cross-encoder reranking, and the replacement was late interaction."}, {"sent_id": 9, "text": "Project Sonata recorded a model choice for reranking on 2026-04-23."}, {"sent_id": 10, "text": "The chosen model was Vela-8B."}, {"sent_id": 11, "text": "On 2026-05-31, Project Nereid retired threshold search in favor of reward reranking."}], "source": "synthetic_rewritten", "title": "Project Update 336"}
{"doc_id": "doc_000337", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Aster's Milestone B1 deadline change on 2026-05-13 replaced 2026-05-24 with 2026-05-19."}, {"sent_id": 3, "text": "For Project Saffron, the run using Vela-7B on Node Fir-10 did not complete on 2026-06-20; the issue was a missing-index error."}, {"sent_id": 4, "text": "The 2026-06-06 method entry for Project Nereid moved from hybrid retrieval to document chunking."}, {"sent_id": 5, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-04-23."}, {"sent_id": 6, "text": "The chosen model was Atlas-7B."}, {"sent_id": 7, "text": "Project Nereid ran Nova-3B on LabQA using Node Aspen-01 on 2026-06-27."}, {"sent_id": 8, "text": "The reported evidence F1 was 0.750."}, {"sent_id": 9, "text": "Project Saffron logged a personnel update on 2026-06-23: Lena Costa took the data steward responsibilities."}, {"sent_id": 10, "text": "Using Node Poplar-12, Helix-7B produced 0.827 accuracy on CedarQA-3 for Project Anchor on 2026-06-22."}, {"sent_id": 11, "text": "Project Saffron recorded a model choice for calibration on 2026-05-07."}, {"sent_id": 12, "text": "The chosen model was Mira-7B."}, {"sent_id": 13, "text": "Project Meridian had a failed run on 2026-04-17 involving Nova-3B and Node Aspen-01."}, {"sent_id": 14, "text": "The recorded cause was a missing-index error."}], "source": "synthetic_rewritten", "title": "Lab Digest 337"}
{"doc_id": "doc_000338", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For reranking, the 2026-04-01 note says Project Aster went with Orchid-7B."}, {"sent_id": 3, "text": "Using Node Maple-01, Marble-3B produced 0.601 latency efficiency score on VestaLogs-3 for Project Saffron on 2026-04-06."}, {"sent_id": 4, "text": "On 2026-05-09, Project Sonata retired query rewriting in favor of sentence pruning."}, {"sent_id": 5, "text": "A compute allocation for Project Sonata was logged on 2026-06-19."}, {"sent_id": 6, "text": "Node Willow-05 supplied 1 GPU."}, {"sent_id": 7, "text": "Project Saffron recorded a model choice for error analysis on 2026-04-29."}, {"sent_id": 8, "text": "The chosen model was Marble-8B."}, {"sent_id": 9, "text": "The 2026-06-28 method entry for Project Meridian moved from BM25 retrieval to evidence pooling."}, {"sent_id": 10, "text": "On 2026-05-17, Project Sonata retired structured prompting in favor of QLoRA adaptation."}], "source": "synthetic_rewritten", "title": "Project Update 338"}
{"doc_id": "doc_000339", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron recorded a model choice for error analysis on 2026-06-21."}, {"sent_id": 3, "text": "The chosen model was Marble-3B."}, {"sent_id": 4, "text": "The staffing note for Project Sonata was dated 2026-06-09."}, {"sent_id": 5, "text": "It named Dr. Jonas Nolan for the retrieval owner duties."}, {"sent_id": 6, "text": "Project Meridian recorded a model choice for calibration on 2026-05-01."}, {"sent_id": 7, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 8, "text": "Project Aster recorded a model choice for claim classification on 2026-04-22."}, {"sent_id": 9, "text": "The chosen model was Helix-3B."}, {"sent_id": 10, "text": "The 2026-04-19 method entry for Project Saffron moved from rank fusion to teacher distillation."}, {"sent_id": 11, "text": "The 2026-06-20 compute note links Node Willow-05 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 12, "text": "On 2026-06-26, the Nested Verifier Study 3 review reported hard-negative mining and said it did not use a reward model."}, {"sent_id": 13, "text": "Finch-7B was the model selected by Project Anchor on 2026-05-28 for the error analysis track."}], "source": "synthetic_rewritten", "title": "Lab Digest 339"}
{"doc_id": "doc_000340", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For reranking, the 2026-04-01 note says Project Anchor went with Orchid-3B."}, {"sent_id": 3, "text": "The 2026-05-16 result for Project Nereid put Cedar-3B on NereidNotes-3 at 0.726 evidence F1, with Node Laurel-11 handling the run."}, {"sent_id": 4, "text": "In the 2026-05-18 staffing section, Project Meridian listed Camila Quinn as the evaluation owner."}, {"sent_id": 5, "text": "In the 2026-05-18 staffing section, Project Sonata listed Mara Quinn as the lead."}, {"sent_id": 6, "text": "The 2026-05-02 compute note links Node Hazel-14 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 7, "text": "The 2026-05-23 method entry for Project Anchor moved from cross-encoder reranking to late interaction."}], "source": "synthetic_rewritten", "title": "Review Packet 340"}
{"doc_id": "doc_000341", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-04-09 failure note says Quartz-8B on Node Pine-07 failed for Project Sonata because of an unstable-validation-loss error."}, {"sent_id": 3, "text": "The staffing note for Project Aster was dated 2026-06-23."}, {"sent_id": 4, "text": "It named Dr. Noah Vale for the lead duties."}, {"sent_id": 5, "text": "For claim classification, the 2026-05-06 note says Project Nereid went with Kestrel-8B."}, {"sent_id": 6, "text": "For error analysis, the 2026-06-23 note says Project Sonata went with Marble-3B."}, {"sent_id": 7, "text": "The 2026-04-04 method entry for Project Anchor moved from metric smoothing to hard-negative mining."}, {"sent_id": 8, "text": "Project Aster's Milestone B1 deadline change on 2026-06-04 replaced 2026-06-25 with 2026-06-30."}, {"sent_id": 9, "text": "Project Sonata ran Mira-7B on TraceEval-3 using Node Sycamore-13 on 2026-04-11."}, {"sent_id": 10, "text": "The reported evidence F1 was 0.628."}, {"sent_id": 11, "text": "For error analysis, the 2026-05-28 note says Project Nereid went with Cedar-3B."}, {"sent_id": 12, "text": "Project Aster recorded a model choice for reranking on 2026-05-03."}, {"sent_id": 13, "text": "The chosen model was Orchid-8B."}], "source": "synthetic_rewritten", "title": "Run Brief 341"}
{"doc_id": "doc_000342", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron recorded a model choice for error analysis on 2026-04-17."}, {"sent_id": 3, "text": "The chosen model was Marble-7B."}, {"sent_id": 4, "text": "Project Aster had a failed run on 2026-06-11 involving Orchid-3B and Node Cedar-02."}, {"sent_id": 5, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 6, "text": "The 2026-05-04 result for Project Saffron put Vela-8B on LumenFacts-3 at 0.565 latency efficiency score, with Node Fir-10 handling the run."}, {"sent_id": 7, "text": "The 2026-04-14 result for Project Meridian put Nova-3B on LumenFacts-2 at 0.795 accuracy, with Node Aspen-01 handling the run."}, {"sent_id": 8, "text": "Using Node Fir-10, Vela-8B produced 0.685 accuracy on LumenFacts-3 for Project Saffron on 2026-05-19."}, {"sent_id": 9, "text": "Using Node Maple-01, Marble-8B produced 0.826 macro F1 on NereidNotes for Project Sonata on 2026-06-30."}], "source": "synthetic_rewritten", "title": "Review Packet 342"}
{"doc_id": "doc_000343", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-05-19 staffing section, Project Anchor listed Leo Hale as the data steward."}, {"sent_id": 3, "text": "For Project Saffron, the run using Quartz-7B on Node Pine-07 did not complete on 2026-05-15; the issue was a missing-index error."}, {"sent_id": 4, "text": "Project Sonata's Milestone D2 deadline change on 2026-06-09 replaced 2026-07-04 with 2026-07-13."}, {"sent_id": 5, "text": "The 2026-04-26 method entry for Project Meridian moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 6, "text": "The staffing note for Project Saffron was dated 2026-06-22."}, {"sent_id": 7, "text": "It named Priya Moss for the data steward duties."}, {"sent_id": 8, "text": "For Project Sonata, the run using Mira-3B on Node Sycamore-13 did not complete on 2026-04-25; the issue was an out-of-memory error."}, {"sent_id": 9, "text": "The 2026-05-23 compute note links Node Laurel-11 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 10, "text": "Project Nereid recorded a model choice for error analysis on 2026-05-24."}, {"sent_id": 11, "text": "The chosen model was Cedar-8B."}], "source": "synthetic_rewritten", "title": "Run Brief 343"}
{"doc_id": "doc_000344", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone V1 for Project Meridian was revised on 2026-04-14."}, {"sent_id": 3, "text": "The schedule moved from 2026-04-30 to 2026-05-08."}, {"sent_id": 4, "text": "In the 2026-04-27 staffing section, Project Saffron listed Petra Gray as the evaluation owner."}, {"sent_id": 5, "text": "Project Saffron logged a personnel update on 2026-05-26: Lena Sol took the evaluation owner responsibilities."}, {"sent_id": 6, "text": "On 2026-05-21, Project Anchor shifted Milestone Z1: 2026-05-29 became 2026-06-06."}, {"sent_id": 7, "text": "On 2026-04-01, Project Meridian retired sentence pruning in favor of cross-encoder reranking."}, {"sent_id": 8, "text": "Atlas-7B was the model selected by Project Meridian on 2026-06-12 for the evidence retrieval track."}], "source": "synthetic_rewritten", "title": "Review Packet 344"}
{"doc_id": "doc_000345", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The staffing note for Project Aster was dated 2026-05-11."}, {"sent_id": 3, "text": "It named Felix Lind for the lead duties."}, {"sent_id": 4, "text": "The 2026-04-15 method entry for Project Aster moved from evidence pooling to calibrated voting."}, {"sent_id": 5, "text": "Project Saffron's Milestone D2 deadline change on 2026-05-13 replaced 2026-06-02 with 2026-06-06."}, {"sent_id": 6, "text": "On 2026-05-12, Project Aster shifted Milestone N1: 2026-05-29 became 2026-06-07."}, {"sent_id": 7, "text": "Project Nereid logged a personnel update on 2026-04-14: Dr. Julian Gray took the lead responsibilities."}, {"sent_id": 8, "text": "Project Anchor received 2 GPUs from Node Spruce-03 in the 2026-04-25 allocation log."}, {"sent_id": 9, "text": "For Project Aster, the run using Orchid-3B on Node Cedar-02 did not complete on 2026-04-23; the issue was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Run Brief 345"}
{"doc_id": "doc_000346", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian logged a personnel update on 2026-05-19: Mira Nolan took the evaluation owner responsibilities."}, {"sent_id": 3, "text": "Project Sonata had a failed run on 2026-04-02 involving Marble-3B and Node Maple-01."}, {"sent_id": 4, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 5, "text": "The 2026-06-20 method entry for Project Aster moved from contrastive tuning to threshold search."}, {"sent_id": 6, "text": "Project Aster received 3 GPUs from Node Cedar-02 in the 2026-04-25 allocation log."}, {"sent_id": 7, "text": "Project Aster logged a personnel update on 2026-05-11: Mina Shah took the lead responsibilities."}, {"sent_id": 8, "text": "Project Sonata's Milestone L1 deadline change on 2026-06-30 replaced 2026-07-21 with 2026-07-26."}, {"sent_id": 9, "text": "Vela-8B was the model selected by Project Sonata on 2026-06-14 for the reranking track."}], "source": "synthetic_rewritten", "title": "Review Packet 346"}
{"doc_id": "doc_000347", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-01 compute note links Node Aspen-01 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 3, "text": "Project Sonata's Milestone F1 deadline change on 2026-05-05 replaced 2026-05-28 with 2026-05-26."}, {"sent_id": 4, "text": "The 2026-06-06 method entry for Project Meridian moved from reward reranking to data mixing."}, {"sent_id": 5, "text": "On 2026-05-24, Project Meridian retired query rewriting in favor of sentence pruning."}, {"sent_id": 6, "text": "The 2026-05-21 compute note links Node Willow-05 to Project Sonata with an allocation of 3 GPUs."}, {"sent_id": 7, "text": "In the 2026-05-05 staffing section, Project Saffron listed Lena Costa as the data steward."}, {"sent_id": 8, "text": "Project Nereid ran Nimbus-3B on TraceEval-2 using Node Elm-08 on 2026-05-11."}, {"sent_id": 9, "text": "The reported evidence F1 was 0.738."}], "source": "synthetic_rewritten", "title": "Run Brief 347"}
{"doc_id": "doc_000348", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron had a method update on 2026-04-19."}, {"sent_id": 3, "text": "The old approach was structured prompting, and the replacement was QLoRA adaptation."}, {"sent_id": 4, "text": "The staffing note for Project Sonata was dated 2026-06-23."}, {"sent_id": 5, "text": "It named Sofia Frost for the lead duties."}, {"sent_id": 6, "text": "Project Aster recorded a model choice for evidence retrieval on 2026-05-17."}, {"sent_id": 7, "text": "The chosen model was River-3B."}, {"sent_id": 8, "text": "Project Nereid logged a personnel update on 2026-05-11: Darian Hale took the retrieval owner responsibilities."}, {"sent_id": 9, "text": "Project Meridian's Milestone P1 deadline change on 2026-04-02 replaced 2026-04-20 with 2026-04-30."}, {"sent_id": 10, "text": "Project Aster had a method update on 2026-06-21."}, {"sent_id": 11, "text": "The old approach was metric smoothing, and the replacement was hard-negative mining."}], "source": "synthetic_rewritten", "title": "Review Packet 348"}
{"doc_id": "doc_000349", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian logged a personnel update on 2026-05-11: Mira Nolan took the evaluation owner responsibilities."}, {"sent_id": 3, "text": "Project Saffron had a failed run on 2026-06-18 involving Marble-3B and Node Maple-01."}, {"sent_id": 4, "text": "The recorded cause was a missing-index error."}, {"sent_id": 5, "text": "On 2026-04-22, Project Aster shifted Milestone B1: 2026-05-13 became 2026-05-18."}, {"sent_id": 6, "text": "For claim classification, the 2026-06-07 note says Project Nereid went with Kestrel-3B."}, {"sent_id": 7, "text": "The 2026-05-16 compute note links Node Juniper-06 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 8, "text": "The 2026-05-31 result for Project Sonata put Quartz-8B on OrionBench at 0.742 macro F1, with Node Pine-07 handling the run."}], "source": "synthetic_rewritten", "title": "Run Brief 349"}
{"doc_id": "doc_000350", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-15, Project Sonata shifted Milestone X1: 2026-04-22 became 2026-04-21."}, {"sent_id": 3, "text": "Project Saffron logged a personnel update on 2026-06-02: Lena Costa took the data steward responsibilities."}, {"sent_id": 4, "text": "The staffing note for Project Aster was dated 2026-06-21."}, {"sent_id": 5, "text": "It named Dr. Noah Chen for the retrieval owner duties."}, {"sent_id": 6, "text": "Milestone J1 for Project Meridian was revised on 2026-06-30."}, {"sent_id": 7, "text": "The schedule moved from 2026-07-10 to 2026-07-20."}, {"sent_id": 8, "text": "The staffing note for Project Anchor was dated 2026-06-09."}, {"sent_id": 9, "text": "It named Iris Stone for the data steward duties."}, {"sent_id": 10, "text": "The staffing note for Project Anchor was dated 2026-06-29."}, {"sent_id": 11, "text": "It named Leo Park for the evaluation owner duties."}, {"sent_id": 12, "text": "Using Node Hazel-14, Atlas-3B produced 0.594 macro F1 on RiverBench-2 for Project Nereid on 2026-05-19."}, {"sent_id": 13, "text": "In the 2026-04-06 staffing section, Project Nereid listed Vera Kim as the lead."}], "source": "synthetic_rewritten", "title": "Lab Digest 350"}
{"doc_id": "doc_000351", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-04-16 method entry for Project Anchor moved from structured prompting to QLoRA adaptation."}, {"sent_id": 3, "text": "The 2026-04-02 failure note says Helix-3B on Node Poplar-12 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 4, "text": "In the 2026-04-14 staffing section, Project Meridian listed Dr. Ravi Reed as the evaluation owner."}, {"sent_id": 5, "text": "Project Meridian received 4 GPUs from Node Laurel-11 in the 2026-06-13 allocation log."}, {"sent_id": 6, "text": "The 2026-04-15 method entry for Project Anchor moved from data mixing to contrastive tuning."}, {"sent_id": 7, "text": "Project Sonata received 1 GPU from Node Pine-07 in the 2026-06-20 allocation log."}, {"sent_id": 8, "text": "In the 2026-05-26 staffing section, Project Meridian listed Camila Quinn as the evaluation owner."}, {"sent_id": 9, "text": "For calibration, the 2026-04-23 note says Project Aster went with Lumen-8B."}], "source": "synthetic_rewritten", "title": "Project Update 351"}
{"doc_id": "doc_000352", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-24 method entry for Project Nereid moved from chain verification to structured prompting."}, {"sent_id": 3, "text": "Quartz-8B was the model selected by Project Saffron on 2026-06-25 for the claim classification track."}, {"sent_id": 4, "text": "On 2026-04-24, Project Aster shifted Milestone Z1: 2026-05-07 became 2026-05-12."}, {"sent_id": 5, "text": "On 2026-06-10, Project Saffron shifted Milestone X1: 2026-06-22 became 2026-06-30."}, {"sent_id": 6, "text": "In the 2026-05-17 staffing section, Project Saffron listed Lena Sol as the evaluation owner."}, {"sent_id": 7, "text": "Milestone R1 for Project Saffron was revised on 2026-05-07."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-31 to 2026-06-08."}, {"sent_id": 9, "text": "Project Meridian had a failed run on 2026-05-22 involving Cedar-8B and Node Laurel-11."}, {"sent_id": 10, "text": "The recorded cause was a missing-index error."}, {"sent_id": 11, "text": "The 2026-05-21 failure note says Lumen-8B on Node Spruce-03 failed for Project Aster because of an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Lab Digest 352"}
{"doc_id": "doc_000353", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Marble-7B was the model selected by Project Saffron on 2026-05-27 for the error analysis track."}, {"sent_id": 3, "text": "Kestrel-8B was the model selected by Project Meridian on 2026-05-10 for the claim classification track."}, {"sent_id": 4, "text": "Project Nereid had a method update on 2026-05-27."}, {"sent_id": 5, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 6, "text": "The 2026-04-29 method entry for Project Aster moved from dense retrieval to rank fusion."}, {"sent_id": 7, "text": "The 2026-04-20 result for Project Meridian put Nova-3B on LumenFacts-2 at 0.765 accuracy, with Node Aspen-01 handling the run."}, {"sent_id": 8, "text": "For Project Meridian, the run using Atlas-8B on Node Hazel-14 did not complete on 2026-06-27; the issue was a missing-index error."}, {"sent_id": 9, "text": "Lumen-8B was the model selected by Project Aster on 2026-05-12 for the calibration track."}, {"sent_id": 10, "text": "Using Node Hazel-14, Atlas-7B produced 0.759 latency efficiency score on SignalSet-3 for Project Meridian on 2026-05-25."}], "source": "synthetic_rewritten", "title": "Project Update 353"}
{"doc_id": "doc_000354", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-06-07 method entry for Project Saffron moved from QLoRA adaptation to hybrid retrieval."}, {"sent_id": 3, "text": "Project Nereid logged a personnel update on 2026-06-08: Evan Moss took the lead responsibilities."}, {"sent_id": 4, "text": "Nova-7B was the model selected by Project Meridian on 2026-06-10 for the reranking track."}, {"sent_id": 5, "text": "Project Anchor's Milestone Z1 deadline change on 2026-05-28 replaced 2026-06-15 with 2026-06-21."}, {"sent_id": 6, "text": "The 2026-06-25 failure note says Vela-8B on Node Fir-10 failed for Project Saffron because of a checkpoint-mismatch error."}, {"sent_id": 7, "text": "Project Meridian had a method update on 2026-05-10."}, {"sent_id": 8, "text": "The old approach was calibrated voting, and the replacement was confidence calibration."}, {"sent_id": 9, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-06-28."}, {"sent_id": 10, "text": "The chosen model was Aster-7B."}], "source": "synthetic_rewritten", "title": "Lab Digest 354"}
{"doc_id": "doc_000355", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron recorded a model choice for reranking on 2026-04-22."}, {"sent_id": 3, "text": "The chosen model was Vela-8B."}, {"sent_id": 4, "text": "The staffing note for Project Aster was dated 2026-04-07."}, {"sent_id": 5, "text": "It named Talia Marin for the lead duties."}, {"sent_id": 6, "text": "Project Aster received 1 GPU from Node Juniper-06 in the 2026-04-09 allocation log."}, {"sent_id": 7, "text": "The 2026-06-10 method entry for Project Meridian moved from LoRA adaptation to alias expansion."}, {"sent_id": 8, "text": "Project Sonata recorded a model choice for reranking on 2026-04-10."}, {"sent_id": 9, "text": "The chosen model was Vela-3B."}, {"sent_id": 10, "text": "Project Aster logged a personnel update on 2026-06-02: Talia Marin took the lead responsibilities."}, {"sent_id": 11, "text": "Using Node Willow-05, Aster-8B produced 0.544 evidence F1 on RiverBench-3 for Project Sonata on 2026-04-20."}], "source": "synthetic_rewritten", "title": "Project Update 355"}
{"doc_id": "doc_000356", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Using Node Hazel-14, Atlas-7B produced 0.564 evidence F1 on RiverBench-2 for Project Nereid on 2026-05-12."}, {"sent_id": 3, "text": "In the 2026-06-23 staffing section, Project Meridian listed Elian Shah as the data steward."}, {"sent_id": 4, "text": "The 2026-05-24 method entry for Project Anchor moved from QLoRA adaptation to hybrid retrieval."}, {"sent_id": 5, "text": "Project Sonata ran Mira-8B on TraceEval-3 using Node Sycamore-13 on 2026-06-01."}, {"sent_id": 6, "text": "The reported evidence F1 was 0.568."}, {"sent_id": 7, "text": "A compute allocation for Project Saffron was logged on 2026-04-09."}, {"sent_id": 8, "text": "Node Willow-05 supplied 2 GPUs."}, {"sent_id": 9, "text": "On 2026-05-31, Project Saffron retired evidence pooling in favor of calibrated voting."}, {"sent_id": 10, "text": "For claim classification, the 2026-05-21 note says Project Meridian went with Kestrel-3B."}], "source": "synthetic_rewritten", "title": "Lab Digest 356"}
{"doc_id": "doc_000357", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Using Node Birch-04, Kestrel-7B produced 0.717 accuracy on CedarQA for Project Meridian on 2026-05-23."}, {"sent_id": 3, "text": "Vela-7B was the model selected by Project Sonata on 2026-04-30 for the reranking track."}, {"sent_id": 4, "text": "Project Saffron ran Vela-7B on LumenFacts-3 using Node Fir-10 on 2026-05-23."}, {"sent_id": 5, "text": "The reported accuracy was 0.535."}, {"sent_id": 6, "text": "River-3B was the model selected by Project Anchor on 2026-04-12 for the evidence retrieval track."}, {"sent_id": 7, "text": "Project Sonata received 1 GPU from Node Sycamore-13 in the 2026-05-09 allocation log."}, {"sent_id": 8, "text": "Project Sonata had a method update on 2026-05-27."}, {"sent_id": 9, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}], "source": "synthetic_rewritten", "title": "Project Update 357"}
{"doc_id": "doc_000358", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Milestone D1 for Project Nereid was revised on 2026-04-24."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-11 to 2026-05-20."}, {"sent_id": 4, "text": "For Project Aster, the run using Lumen-3B on Node Spruce-03 did not complete on 2026-06-25; the issue was an out-of-memory error."}, {"sent_id": 5, "text": "For Project Nereid, the run using Kestrel-3B on Node Birch-04 did not complete on 2026-06-25; the issue was an out-of-memory error."}, {"sent_id": 6, "text": "Project Anchor had a method update on 2026-06-28."}, {"sent_id": 7, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 8, "text": "In the 2026-05-25 staffing section, Project Aster listed Felix Lind as the lead."}, {"sent_id": 9, "text": "In the 2026-05-19 staffing section, Project Sonata listed Mara Lane as the retrieval owner."}, {"sent_id": 10, "text": "Project Meridian's Milestone V1 deadline change on 2026-04-13 replaced 2026-05-09 with 2026-05-15."}], "source": "synthetic_rewritten", "title": "Lab Digest 358"}
{"doc_id": "doc_000359", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-06-25, Project Sonata shifted Milestone R1: 2026-07-14 became 2026-07-11."}, {"sent_id": 3, "text": "In the 2026-04-05 staffing section, Project Saffron listed Dr. Theo Lind as the evaluation owner."}, {"sent_id": 4, "text": "Project Nereid had a method update on 2026-04-18."}, {"sent_id": 5, "text": "The old approach was alias expansion, and the replacement was dense retrieval."}, {"sent_id": 6, "text": "Project Anchor logged a personnel update on 2026-06-23: Dr. Arun Bauer took the evaluation owner responsibilities."}, {"sent_id": 7, "text": "For reranking, the 2026-04-05 note says Project Sonata went with Vela-8B."}, {"sent_id": 8, "text": "Project Nereid had a failed run on 2026-04-09 involving Atlas-3B and Node Hazel-14."}, {"sent_id": 9, "text": "The recorded cause was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Project Update 359"}
{"doc_id": "doc_000360", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron ran Mira-7B on MemoTrace-2 using Node Sycamore-13 on 2026-04-13."}, {"sent_id": 3, "text": "The reported latency efficiency score was 0.553."}, {"sent_id": 4, "text": "The 2026-04-06 compute note links Node Aspen-01 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 5, "text": "Helix-7B was the model selected by Project Aster on 2026-06-03 for the claim classification track."}, {"sent_id": 6, "text": "The staffing note for Project Aster was dated 2026-05-19."}, {"sent_id": 7, "text": "It named Felix Lind for the lead duties."}, {"sent_id": 8, "text": "Project Saffron ran Aster-3B on SignalSet using Node Willow-05 on 2026-06-02."}, {"sent_id": 9, "text": "The reported accuracy was 0.589."}, {"sent_id": 10, "text": "On 2026-04-11, Project Anchor retired structured prompting in favor of QLoRA adaptation."}, {"sent_id": 11, "text": "Project Meridian logged a personnel update on 2026-05-12: Mira Sato took the data steward responsibilities."}, {"sent_id": 12, "text": "Finch-7B was the model selected by Project Anchor on 2026-05-22 for the error analysis track."}], "source": "synthetic_rewritten", "title": "Run Brief 360"}
{"doc_id": "doc_000361", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-09 failure note says Helix-3B on Node Poplar-12 failed for Project Anchor because of a checkpoint-mismatch error."}, {"sent_id": 3, "text": "Helix-7B was the model selected by Project Aster on 2026-06-21 for the claim classification track."}, {"sent_id": 4, "text": "Project Sonata logged a personnel update on 2026-04-13: Mara Quinn took the lead responsibilities."}, {"sent_id": 5, "text": "Project Nereid ran Kestrel-3B on OrionBench-3 using Node Birch-04 on 2026-04-04."}, {"sent_id": 6, "text": "The reported evidence F1 was 0.702."}, {"sent_id": 7, "text": "On 2026-05-20, Project Nereid shifted Milestone V1: 2026-05-31 became 2026-05-26."}, {"sent_id": 8, "text": "Project Anchor recorded a model choice for reranking on 2026-05-20."}, {"sent_id": 9, "text": "The chosen model was Orchid-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 361"}
{"doc_id": "doc_000362", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor had a failed run on 2026-06-18 involving Finch-8B and Node Juniper-06."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "In the 2026-05-05 staffing section, Project Aster listed Mina Shah as the lead."}, {"sent_id": 5, "text": "The 2026-06-25 method entry for Project Sonata moved from QLoRA adaptation to hybrid retrieval."}, {"sent_id": 6, "text": "The 2026-06-27 method entry for Project Meridian moved from hard-negative mining to LoRA adaptation."}, {"sent_id": 7, "text": "On 2026-06-24, Project Meridian shifted Milestone J1: 2026-07-14 became 2026-07-18."}, {"sent_id": 8, "text": "For Project Aster, the run using River-3B on Node Rowan-09 did not complete on 2026-06-04; the issue was an unstable-validation-loss error."}, {"sent_id": 9, "text": "The 2026-04-11 compute note links Node Birch-04 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 10, "text": "Milestone F1 for Project Saffron was revised on 2026-06-22."}, {"sent_id": 11, "text": "The schedule moved from 2026-06-30 to 2026-07-04."}], "source": "synthetic_rewritten", "title": "Run Brief 362"}
{"doc_id": "doc_000363", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Aster was logged on 2026-06-29."}, {"sent_id": 3, "text": "Node Poplar-12 supplied 1 GPU."}, {"sent_id": 4, "text": "The 2026-05-02 method entry for Project Sonata moved from chain verification to structured prompting."}, {"sent_id": 5, "text": "On 2026-06-14, Project Sonata retired alias expansion in favor of dense retrieval."}, {"sent_id": 6, "text": "Milestone X1 for Project Saffron was revised on 2026-05-27."}, {"sent_id": 7, "text": "The schedule moved from 2026-06-08 to 2026-06-12."}, {"sent_id": 8, "text": "For claim classification, the 2026-05-17 note says Project Aster went with Helix-8B."}, {"sent_id": 9, "text": "Project Meridian recorded a model choice for calibration on 2026-04-10."}, {"sent_id": 10, "text": "The chosen model was Nimbus-8B."}, {"sent_id": 11, "text": "The 2026-06-23 result for Project Saffron put Vela-7B on LumenFacts-3 at 0.835 accuracy, with Node Fir-10 handling the run."}], "source": "synthetic_rewritten", "title": "Review Packet 363"}
{"doc_id": "doc_000364", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Finch-8B was the model selected by Project Anchor on 2026-04-15 for the error analysis track."}, {"sent_id": 3, "text": "Project Saffron received 4 GPUs from Node Sycamore-13 in the 2026-06-27 allocation log."}, {"sent_id": 4, "text": "Milestone X1 for Project Saffron was revised on 2026-06-17."}, {"sent_id": 5, "text": "The schedule moved from 2026-07-09 to 2026-07-15."}, {"sent_id": 6, "text": "Project Anchor logged a personnel update on 2026-04-20: Kira Frost took the data steward responsibilities."}, {"sent_id": 7, "text": "Using Node Sycamore-13, Mira-8B produced 0.853 latency efficiency score on MemoTrace-2 for Project Saffron on 2026-06-09."}, {"sent_id": 8, "text": "The 2026-05-18 result for Project Saffron put Quartz-7B on CedarQA-2 at 0.547 accuracy, with Node Pine-07 handling the run."}, {"sent_id": 9, "text": "Project Anchor ran Finch-7B on VestaLogs using Node Juniper-06 on 2026-04-28."}, {"sent_id": 10, "text": "The reported accuracy was 0.881."}, {"sent_id": 11, "text": "On 2026-05-17, Project Meridian retired confidence calibration in favor of chain verification."}, {"sent_id": 12, "text": "Using Node Fir-10, Vela-3B produced 0.565 latency efficiency score on LumenFacts-3 for Project Saffron on 2026-05-30."}], "source": "synthetic_rewritten", "title": "Run Brief 364"}
{"doc_id": "doc_000365", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-06-14, Project Nereid retired structured prompting in favor of QLoRA adaptation."}, {"sent_id": 3, "text": "Project Anchor's Milestone H1 deadline change on 2026-06-18 replaced 2026-07-02 with 2026-07-12."}, {"sent_id": 4, "text": "Using Node Cedar-02, Orchid-7B produced 0.800 evidence F1 on LabQA-3 for Project Aster on 2026-04-13."}, {"sent_id": 5, "text": "On 2026-06-24, Project Anchor retired document chunking in favor of metric smoothing."}, {"sent_id": 6, "text": "Project Saffron had a method update on 2026-05-24."}, {"sent_id": 7, "text": "The old approach was structured prompting, and the replacement was QLoRA adaptation."}, {"sent_id": 8, "text": "Milestone B2 for Project Nereid was revised on 2026-04-29."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-08 to 2026-05-13."}], "source": "synthetic_rewritten", "title": "Review Packet 365"}
{"doc_id": "doc_000366", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-04-01, Project Aster retired LoRA adaptation in favor of alias expansion."}, {"sent_id": 3, "text": "In the 2026-04-06 staffing section, Project Meridian listed Mira Sato as the data steward."}, {"sent_id": 4, "text": "The review note for Amber Ranking Study 3 was dated 2026-04-08."}, {"sent_id": 5, "text": "It described the method as hybrid retrieval and said it did not use a reward model."}, {"sent_id": 6, "text": "The staffing note for Project Meridian was dated 2026-06-22."}, {"sent_id": 7, "text": "It named Dr. Ravi Nadir for the data steward duties."}, {"sent_id": 8, "text": "The 2026-06-06 result for Project Aster put Helix-3B on OrionBench-2 at 0.842 macro F1, with Node Poplar-12 handling the run."}, {"sent_id": 9, "text": "The 2026-04-23 compute note links Node Hazel-14 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 10, "text": "Milestone L1 for Project Saffron was revised on 2026-06-25."}, {"sent_id": 11, "text": "The schedule moved from 2026-07-21 to 2026-07-27."}, {"sent_id": 12, "text": "The 2026-04-14 result for Project Meridian put Kestrel-7B on CedarQA at 0.717 latency efficiency score, with Node Birch-04 handling the run."}, {"sent_id": 13, "text": "A compute allocation for Project Nereid was logged on 2026-06-27."}, {"sent_id": 14, "text": "Node Laurel-11 supplied 1 GPU."}], "source": "synthetic_rewritten", "title": "Run Brief 366"}
{"doc_id": "doc_000367", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Milestone X1 for Project Sonata was revised on 2026-04-22."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-09 to 2026-05-18."}, {"sent_id": 4, "text": "On 2026-06-13, Project Nereid retired LoRA adaptation in favor of alias expansion."}, {"sent_id": 5, "text": "In the 2026-04-28 staffing section, Project Meridian listed Camila Brooks as the data steward."}, {"sent_id": 6, "text": "Project Nereid logged a personnel update on 2026-04-28: Dr. Julian Gray took the lead responsibilities."}, {"sent_id": 7, "text": "For Project Sonata, the run using Aster-7B on Node Willow-05 did not complete on 2026-06-18; the issue was an unstable-validation-loss error."}, {"sent_id": 8, "text": "Project Sonata's Milestone L1 deadline change on 2026-04-08 replaced 2026-04-19 with 2026-04-14."}, {"sent_id": 9, "text": "The 2026-06-05 compute note links Node Rowan-09 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 10, "text": "For Project Nereid, the run using Atlas-7B on Node Hazel-14 did not complete on 2026-06-26; the issue was an out-of-memory error."}, {"sent_id": 11, "text": "For Project Aster, the run using Lumen-3B on Node Spruce-03 did not complete on 2026-04-10; the issue was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Review Packet 367"}
{"doc_id": "doc_000368", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian logged a personnel update on 2026-06-01: Camila Quinn took the evaluation owner responsibilities."}, {"sent_id": 3, "text": "Project Anchor's Milestone T1 deadline change on 2026-06-26 replaced 2026-07-16 with 2026-07-20."}, {"sent_id": 4, "text": "In the 2026-05-04 staffing section, Project Sonata listed Rohan Singh as the retrieval owner."}, {"sent_id": 5, "text": "For evidence retrieval, the 2026-04-19 note says Project Sonata went with Aster-3B."}, {"sent_id": 6, "text": "Project Sonata had a failed run on 2026-04-16 involving Quartz-7B and Node Pine-07."}, {"sent_id": 7, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 8, "text": "Project Nereid logged a personnel update on 2026-04-21: Dr. Julian Stone took the retrieval owner responsibilities."}], "source": "synthetic_rewritten", "title": "Run Brief 368"}
{"doc_id": "doc_000369", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-06-10, Project Aster retired temporal filtering in favor of alias expansion."}, {"sent_id": 3, "text": "The 2026-04-17 compute note links Node Pine-07 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 4, "text": "Project Saffron's Milestone D2 deadline change on 2026-06-05 replaced 2026-06-25 with 2026-06-29."}, {"sent_id": 5, "text": "In the 2026-04-28 staffing section, Project Nereid listed Evan Iyer as the retrieval owner."}, {"sent_id": 6, "text": "The 2026-04-04 compute note links Node Spruce-03 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 7, "text": "The 2026-05-22 failure note says Helix-3B on Node Poplar-12 failed for Project Aster because of an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Review Packet 369"}
{"doc_id": "doc_000370", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Using Node Elm-08, Nimbus-7B produced 0.738 evidence F1 on TraceEval-2 for Project Nereid on 2026-06-06."}, {"sent_id": 3, "text": "Project Saffron logged a personnel update on 2026-04-27: Dr. Theo Grant took the data steward responsibilities."}, {"sent_id": 4, "text": "A compute allocation for Project Saffron was logged on 2026-05-23."}, {"sent_id": 5, "text": "Node Willow-05 supplied 2 GPUs."}, {"sent_id": 6, "text": "On 2026-04-26, Project Meridian retired QLoRA adaptation in favor of hybrid retrieval."}, {"sent_id": 7, "text": "For reranking, the 2026-04-26 note says Project Aster went with Orchid-3B."}, {"sent_id": 8, "text": "Project Nereid received 1 GPU from Node Laurel-11 in the 2026-06-13 allocation log."}, {"sent_id": 9, "text": "Project Anchor received 4 GPUs from Node Juniper-06 in the 2026-06-27 allocation log."}, {"sent_id": 10, "text": "For evidence retrieval, the 2026-05-05 note says Project Nereid went with Atlas-8B."}, {"sent_id": 11, "text": "On 2026-05-09, Project Nereid retired QLoRA adaptation in favor of hybrid retrieval."}], "source": "synthetic_rewritten", "title": "Project Update 370"}
{"doc_id": "doc_000371", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For Project Meridian, the run using Kestrel-7B on Node Birch-04 did not complete on 2026-06-11; the issue was a checkpoint-mismatch error."}, {"sent_id": 3, "text": "The 2026-04-08 method entry for Project Anchor moved from alias expansion to dense retrieval."}, {"sent_id": 4, "text": "Nova-3B was the model selected by Project Nereid on 2026-06-18 for the reranking track."}, {"sent_id": 5, "text": "For claim classification, the 2026-06-24 note says Project Meridian went with Kestrel-3B."}, {"sent_id": 6, "text": "In the 2026-06-15 staffing section, Project Saffron listed Petra Gray as the evaluation owner."}, {"sent_id": 7, "text": "Project Nereid had a method update on 2026-06-21."}, {"sent_id": 8, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 9, "text": "Project Nereid received 3 GPUs from Node Laurel-11 in the 2026-05-15 allocation log."}, {"sent_id": 10, "text": "On 2026-06-24, Project Meridian shifted Milestone B2: 2026-07-08 became 2026-07-18."}], "source": "synthetic_rewritten", "title": "Lab Digest 371"}
{"doc_id": "doc_000372", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For error analysis, the 2026-05-27 note says Project Anchor went with Finch-3B."}, {"sent_id": 3, "text": "Project Sonata had a method update on 2026-06-28."}, {"sent_id": 4, "text": "The old approach was late interaction, and the replacement was BM25 retrieval."}, {"sent_id": 5, "text": "Project Anchor recorded a model choice for calibration on 2026-06-25."}, {"sent_id": 6, "text": "The chosen model was Lumen-7B."}, {"sent_id": 7, "text": "Project Sonata had a method update on 2026-05-13."}, {"sent_id": 8, "text": "The old approach was contrastive tuning, and the replacement was threshold search."}, {"sent_id": 9, "text": "Milestone P1 for Project Nereid was revised on 2026-06-30."}, {"sent_id": 10, "text": "The schedule moved from 2026-07-23 to 2026-07-21."}, {"sent_id": 11, "text": "Project Aster had a failed run on 2026-06-12 involving River-7B and Node Rowan-09."}, {"sent_id": 12, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 13, "text": "In the 2026-05-18 staffing section, Project Aster listed Mina Adler as the retrieval owner."}, {"sent_id": 14, "text": "Project Nereid received 1 GPU from Node Hazel-14 in the 2026-05-23 allocation log."}], "source": "synthetic_rewritten", "title": "Project Update 372"}
{"doc_id": "doc_000373", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Meridian logged a personnel update on 2026-04-06: Nadia Chen took the evaluation owner responsibilities."}, {"sent_id": 3, "text": "Project Saffron received 2 GPUs from Node Sycamore-13 in the 2026-06-20 allocation log."}, {"sent_id": 4, "text": "The staffing note for Project Anchor was dated 2026-06-29."}, {"sent_id": 5, "text": "It named Kira Frost for the data steward duties."}, {"sent_id": 6, "text": "Nimbus-3B was the model selected by Project Meridian on 2026-04-30 for the calibration track."}, {"sent_id": 7, "text": "The staffing note for Project Meridian was dated 2026-05-26."}, {"sent_id": 8, "text": "It named Mira Sato for the data steward duties."}, {"sent_id": 9, "text": "Project Anchor received 2 GPUs from Node Cedar-02 in the 2026-04-22 allocation log."}, {"sent_id": 10, "text": "The 2026-04-02 failure note says Nova-3B on Node Aspen-01 failed for Project Meridian because of a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Lab Digest 373"}
{"doc_id": "doc_000374", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Helix-8B was the model selected by Project Aster on 2026-06-19 for the claim classification track."}, {"sent_id": 3, "text": "Project Aster recorded a model choice for error analysis on 2026-04-16."}, {"sent_id": 4, "text": "The chosen model was Finch-8B."}, {"sent_id": 5, "text": "Project Sonata recorded a model choice for claim classification on 2026-04-19."}, {"sent_id": 6, "text": "The chosen model was Quartz-7B."}, {"sent_id": 7, "text": "On 2026-04-02, Project Anchor shifted Milestone T1: 2026-04-22 became 2026-04-30."}, {"sent_id": 8, "text": "Project Saffron had a method update on 2026-05-10."}, {"sent_id": 9, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 10, "text": "Helix-3B was the model selected by Project Anchor on 2026-04-05 for the claim classification track."}, {"sent_id": 11, "text": "Project Meridian had a method update on 2026-06-24."}, {"sent_id": 12, "text": "The old approach was late interaction, and the replacement was BM25 retrieval."}, {"sent_id": 13, "text": "Project Saffron recorded a model choice for claim classification on 2026-06-07."}, {"sent_id": 14, "text": "The chosen model was Quartz-8B."}, {"sent_id": 15, "text": "Project Anchor recorded a model choice for evidence retrieval on 2026-06-17."}, {"sent_id": 16, "text": "The chosen model was River-7B."}], "source": "synthetic_rewritten", "title": "Project Update 374"}
{"doc_id": "doc_000375", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-05-04 staffing section, Project Aster listed Mina Adler as the retrieval owner."}, {"sent_id": 3, "text": "In the 2026-06-28 staffing section, Project Sonata listed Rohan Singh as the retrieval owner."}, {"sent_id": 4, "text": "Project Sonata ran Aster-7B on RiverBench-3 using Node Willow-05 on 2026-04-21."}, {"sent_id": 5, "text": "The reported macro F1 was 0.604."}, {"sent_id": 6, "text": "A compute allocation for Project Saffron was logged on 2026-05-09."}, {"sent_id": 7, "text": "Node Willow-05 supplied 2 GPUs."}, {"sent_id": 8, "text": "Mira-8B was the model selected by Project Saffron on 2026-04-12 for the calibration track."}, {"sent_id": 9, "text": "Milestone H1 for Project Aster was revised on 2026-04-01."}, {"sent_id": 10, "text": "The schedule moved from 2026-04-20 to 2026-04-17."}, {"sent_id": 11, "text": "Orchid-7B was the model selected by Project Aster on 2026-04-19 for the reranking track."}, {"sent_id": 12, "text": "River-7B was the model selected by Project Aster on 2026-04-07 for the evidence retrieval track."}, {"sent_id": 13, "text": "In the 2026-06-21 staffing section, Project Nereid listed Evan Moss as the lead."}], "source": "synthetic_rewritten", "title": "Lab Digest 375"}
{"doc_id": "doc_000376", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-04-02 failure note says Atlas-7B on Node Hazel-14 failed for Project Nereid because of an out-of-memory error."}, {"sent_id": 3, "text": "Project Anchor recorded a model choice for calibration on 2026-06-26."}, {"sent_id": 4, "text": "The chosen model was Lumen-3B."}, {"sent_id": 5, "text": "For Project Sonata, the run using Quartz-3B on Node Pine-07 did not complete on 2026-04-17; the issue was an unstable-validation-loss error."}, {"sent_id": 6, "text": "The 2026-05-31 method entry for Project Meridian moved from document chunking to metric smoothing."}, {"sent_id": 7, "text": "The 2026-06-27 compute note links Node Cedar-02 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 8, "text": "Quartz-7B was the model selected by Project Sonata on 2026-05-22 for the claim classification track."}], "source": "synthetic_rewritten", "title": "Project Update 376"}
{"doc_id": "doc_000377", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-06-08 result for Project Meridian put Nimbus-7B on MemoTrace at 0.663 latency efficiency score, with Node Elm-08 handling the run."}, {"sent_id": 3, "text": "The 2026-05-30 compute note links Node Spruce-03 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 4, "text": "Project Aster received 1 GPU from Node Spruce-03 in the 2026-06-27 allocation log."}, {"sent_id": 5, "text": "Mira-3B was the model selected by Project Sonata on 2026-04-01 for the calibration track."}, {"sent_id": 6, "text": "In the 2026-04-21 staffing section, Project Anchor listed Leo Park as the evaluation owner."}, {"sent_id": 7, "text": "The staffing note for Project Saffron was dated 2026-06-23."}, {"sent_id": 8, "text": "It named Priya Vale for the evaluation owner duties."}, {"sent_id": 9, "text": "Using Node Willow-05, Aster-7B produced 0.529 latency efficiency score on SignalSet for Project Saffron on 2026-06-01."}, {"sent_id": 10, "text": "Using Node Birch-04, Kestrel-8B produced 0.882 evidence F1 on OrionBench-3 for Project Nereid on 2026-05-03."}, {"sent_id": 11, "text": "On 2026-05-20, Project Saffron retired BM25 retrieval in favor of evidence pooling."}], "source": "synthetic_rewritten", "title": "Lab Digest 377"}
{"doc_id": "doc_000378", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid logged a personnel update on 2026-06-30: Nora Sol took the retrieval owner responsibilities."}, {"sent_id": 3, "text": "For calibration, the 2026-05-31 note says Project Sonata went with Mira-3B."}, {"sent_id": 4, "text": "The staffing note for Project Saffron was dated 2026-05-05."}, {"sent_id": 5, "text": "It named Priya Moss for the data steward duties."}, {"sent_id": 6, "text": "A compute allocation for Project Saffron was logged on 2026-05-09."}, {"sent_id": 7, "text": "Node Fir-10 supplied 2 GPUs."}, {"sent_id": 8, "text": "The review note for Stable Chains Study 1 was dated 2026-05-11."}, {"sent_id": 9, "text": "It described the method as document chunking and said it did not use a reward model."}, {"sent_id": 10, "text": "The 2026-04-04 result for Project Saffron put Quartz-8B on CedarQA-2 at 0.637 latency efficiency score, with Node Pine-07 handling the run."}, {"sent_id": 11, "text": "The 2026-04-08 method entry for Project Saffron moved from hard-negative mining to LoRA adaptation."}, {"sent_id": 12, "text": "On 2026-06-18, Project Aster retired chain verification in favor of structured prompting."}], "source": "synthetic_rewritten", "title": "Project Update 378"}
{"doc_id": "doc_000379", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Atlas-8B was the model selected by Project Meridian on 2026-06-03 for the evidence retrieval track."}, {"sent_id": 3, "text": "A compute allocation for Project Anchor was logged on 2026-06-06."}, {"sent_id": 4, "text": "Node Cedar-02 supplied 4 GPUs."}, {"sent_id": 5, "text": "Project Nereid's Milestone D1 deadline change on 2026-04-23 replaced 2026-04-30 with 2026-04-29."}, {"sent_id": 6, "text": "Cedar-7B was the model selected by Project Nereid on 2026-05-19 for the error analysis track."}, {"sent_id": 7, "text": "A compute allocation for Project Sonata was logged on 2026-04-11."}, {"sent_id": 8, "text": "Node Maple-01 supplied 3 GPUs."}, {"sent_id": 9, "text": "The staffing note for Project Nereid was dated 2026-05-26."}, {"sent_id": 10, "text": "It named Darian Grant for the lead duties."}, {"sent_id": 11, "text": "Project Sonata recorded a model choice for evidence retrieval on 2026-05-03."}, {"sent_id": 12, "text": "The chosen model was Aster-7B."}], "source": "synthetic_rewritten", "title": "Lab Digest 379"}
{"doc_id": "doc_000380", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-06-03, Project Sonata shifted Milestone R1: 2026-06-12 became 2026-06-21."}, {"sent_id": 3, "text": "Project Nereid received 1 GPU from Node Laurel-11 in the 2026-04-25 allocation log."}, {"sent_id": 4, "text": "The staffing note for Project Meridian was dated 2026-04-20."}, {"sent_id": 5, "text": "It named Camila Brooks for the data steward duties."}, {"sent_id": 6, "text": "On 2026-06-24, Project Sonata shifted Milestone D2: 2026-07-19 became 2026-07-24."}, {"sent_id": 7, "text": "The 2026-06-05 failure note says Nova-7B on Node Aspen-01 failed for Project Nereid because of an unstable-validation-loss error."}, {"sent_id": 8, "text": "Project Aster logged a personnel update on 2026-04-28: Felix Lind took the lead responsibilities."}, {"sent_id": 9, "text": "Project Aster had a method update on 2026-04-05."}, {"sent_id": 10, "text": "The old approach was temporal filtering, and the replacement was alias expansion."}, {"sent_id": 11, "text": "The staffing note for Project Anchor was dated 2026-04-27."}, {"sent_id": 12, "text": "It named Iris Stone for the data steward duties."}, {"sent_id": 13, "text": "River-3B was the model selected by Project Aster on 2026-04-10 for the evidence retrieval track."}], "source": "synthetic_rewritten", "title": "Review Packet 380"}
{"doc_id": "doc_000381", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-02 compute note links Node Sycamore-13 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 3, "text": "Project Meridian ran Cedar-3B on VestaLogs-2 using Node Laurel-11 on 2026-06-02."}, {"sent_id": 4, "text": "The reported accuracy was 0.771."}, {"sent_id": 5, "text": "For Project Nereid, the run using Atlas-3B on Node Hazel-14 did not complete on 2026-06-18; the issue was an out-of-memory error."}, {"sent_id": 6, "text": "The staffing note for Project Aster was dated 2026-04-07."}, {"sent_id": 7, "text": "It named Anika Costa for the retrieval owner duties."}, {"sent_id": 8, "text": "The 2026-06-14 method entry for Project Meridian moved from temporal filtering to alias expansion."}, {"sent_id": 9, "text": "In the 2026-05-04 staffing section, Project Meridian listed Elian Ames as the evaluation owner."}, {"sent_id": 10, "text": "For error analysis, the 2026-04-01 note says Project Anchor went with Finch-7B."}, {"sent_id": 11, "text": "Nova-7B was the model selected by Project Nereid on 2026-06-11 for the reranking track."}], "source": "synthetic_rewritten", "title": "Run Brief 381"}
{"doc_id": "doc_000382", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Nimbus-3B was the model selected by Project Meridian on 2026-06-21 for the calibration track."}, {"sent_id": 3, "text": "Project Sonata had a failed run on 2026-05-08 involving Marble-3B and Node Maple-01."}, {"sent_id": 4, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 5, "text": "On 2026-04-29, Project Nereid retired hard-negative mining in favor of LoRA adaptation."}, {"sent_id": 6, "text": "A compute allocation for Project Sonata was logged on 2026-05-02."}, {"sent_id": 7, "text": "Node Willow-05 supplied 3 GPUs."}, {"sent_id": 8, "text": "Project Anchor recorded a model choice for calibration on 2026-05-17."}, {"sent_id": 9, "text": "The chosen model was Lumen-7B."}, {"sent_id": 10, "text": "The 2026-05-07 failure note says Aster-3B on Node Willow-05 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 11, "text": "Nested Verifier Study 4 appeared in a 2026-06-17 review entry using rank fusion; the note said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Review Packet 382"}
{"doc_id": "doc_000383", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Aster had a method update on 2026-05-10."}, {"sent_id": 3, "text": "The old approach was cross-encoder reranking, and the replacement was late interaction."}, {"sent_id": 4, "text": "Project Sonata's Milestone X1 deadline change on 2026-05-21 replaced 2026-06-07 with 2026-06-12."}, {"sent_id": 5, "text": "Project Nereid logged a personnel update on 2026-05-04: Dr. Julian Gray took the lead responsibilities."}, {"sent_id": 6, "text": "Project Meridian recorded a model choice for calibration on 2026-04-24."}, {"sent_id": 7, "text": "The chosen model was Nimbus-3B."}, {"sent_id": 8, "text": "Project Anchor's Milestone H1 deadline change on 2026-06-25 replaced 2026-07-19 with 2026-07-27."}, {"sent_id": 9, "text": "On 2026-06-10, Project Sonata retired confidence calibration in favor of chain verification."}, {"sent_id": 10, "text": "On 2026-05-29, Project Anchor shifted Milestone Z1: 2026-06-06 became 2026-06-14."}], "source": "synthetic_rewritten", "title": "Run Brief 383"}
{"doc_id": "doc_000384", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-24, Project Aster shifted Milestone H1: 2026-05-13 became 2026-05-10."}, {"sent_id": 3, "text": "In the 2026-05-31 staffing section, Project Anchor listed Selene Kim as the evaluation owner."}, {"sent_id": 4, "text": "The 2026-04-28 result for Project Aster put Finch-3B on NereidNotes-2 at 0.686 evidence F1, with Node Juniper-06 handling the run."}, {"sent_id": 5, "text": "The staffing note for Project Aster was dated 2026-06-01."}, {"sent_id": 6, "text": "It named Anika Sato for the lead duties."}, {"sent_id": 7, "text": "The staffing note for Project Nereid was dated 2026-06-15."}, {"sent_id": 8, "text": "It named Vera Kim for the lead duties."}, {"sent_id": 9, "text": "For reranking, the 2026-06-05 note says Project Saffron went with Vela-3B."}, {"sent_id": 10, "text": "Project Nereid received 3 GPUs from Node Aspen-01 in the 2026-04-08 allocation log."}, {"sent_id": 11, "text": "For claim classification, the 2026-05-13 note says Project Nereid went with Kestrel-7B."}, {"sent_id": 12, "text": "Project Meridian recorded a model choice for evidence retrieval on 2026-06-23."}, {"sent_id": 13, "text": "The chosen model was Atlas-3B."}], "source": "synthetic_rewritten", "title": "Review Packet 384"}
{"doc_id": "doc_000385", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The staffing note for Project Aster was dated 2026-06-09."}, {"sent_id": 3, "text": "It named Talia Reed for the retrieval owner duties."}, {"sent_id": 4, "text": "Marble-7B was the model selected by Project Saffron on 2026-05-12 for the error analysis track."}, {"sent_id": 5, "text": "Project Sonata's Milestone D2 deadline change on 2026-06-17 replaced 2026-07-02 with 2026-06-28."}, {"sent_id": 6, "text": "Project Nereid recorded a model choice for evidence retrieval on 2026-05-01."}, {"sent_id": 7, "text": "The chosen model was Atlas-8B."}, {"sent_id": 8, "text": "Using Node Spruce-03, Lumen-7B produced 0.803 accuracy on MemoTrace-3 for Project Anchor on 2026-05-11."}, {"sent_id": 9, "text": "The 2026-05-02 result for Project Anchor put Helix-3B on CedarQA-3 at 0.887 accuracy, with Node Poplar-12 handling the run."}, {"sent_id": 10, "text": "The 2026-05-25 result for Project Anchor put Lumen-8B on MemoTrace-3 at 0.863 accuracy, with Node Spruce-03 handling the run."}], "source": "synthetic_rewritten", "title": "Run Brief 385"}
{"doc_id": "doc_000386", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Sonata had a failed run on 2026-04-16 involving Mira-3B and Node Sycamore-13."}, {"sent_id": 3, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 4, "text": "Project Anchor ran River-3B on SignalSet-2 using Node Rowan-09 on 2026-05-31."}, {"sent_id": 5, "text": "The reported latency efficiency score was 0.599."}, {"sent_id": 6, "text": "On 2026-06-20, Project Anchor retired confidence calibration in favor of chain verification."}, {"sent_id": 7, "text": "On 2026-04-29, Project Meridian retired query rewriting in favor of sentence pruning."}, {"sent_id": 8, "text": "Orchid-3B was the model selected by Project Aster on 2026-06-04 for the reranking track."}, {"sent_id": 9, "text": "Project Nereid had a method update on 2026-06-13."}, {"sent_id": 10, "text": "The old approach was metric smoothing, and the replacement was hard-negative mining."}], "source": "synthetic_rewritten", "title": "Review Packet 386"}
{"doc_id": "doc_000387", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "A compute allocation for Project Saffron was logged on 2026-06-19."}, {"sent_id": 3, "text": "Node Fir-10 supplied 2 GPUs."}, {"sent_id": 4, "text": "The 2026-04-05 method entry for Project Aster moved from calibrated voting to confidence calibration."}, {"sent_id": 5, "text": "The 2026-06-07 method entry for Project Anchor moved from query rewriting to sentence pruning."}, {"sent_id": 6, "text": "Project Aster received 3 GPUs from Node Poplar-12 in the 2026-05-02 allocation log."}, {"sent_id": 7, "text": "The 2026-05-18 compute note links Node Hazel-14 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 8, "text": "Project Saffron received 2 GPUs from Node Sycamore-13 in the 2026-04-04 allocation log."}, {"sent_id": 9, "text": "The 2026-05-06 method entry for Project Nereid moved from hybrid retrieval to document chunking."}, {"sent_id": 10, "text": "The 2026-05-09 result for Project Nereid put Atlas-3B on RiverBench-2 at 0.774 evidence F1, with Node Hazel-14 handling the run."}, {"sent_id": 11, "text": "On 2026-06-12, the Stable Chains Study 2 review reported document chunking and said it did not use a reward model."}], "source": "synthetic_rewritten", "title": "Run Brief 387"}
{"doc_id": "doc_000388", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For claim classification, the 2026-06-30 note says Project Saffron went with Quartz-7B."}, {"sent_id": 3, "text": "Project Nereid had a failed run on 2026-04-23 involving Nimbus-3B and Node Elm-08."}, {"sent_id": 4, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 5, "text": "Project Aster ran River-7B on RiverBench using Node Rowan-09 on 2026-05-25."}, {"sent_id": 6, "text": "The reported evidence F1 was 0.824."}, {"sent_id": 7, "text": "In the 2026-05-10 staffing section, Project Nereid listed Nora Bauer as the lead."}, {"sent_id": 8, "text": "Milestone V1 for Project Meridian was revised on 2026-04-15."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-01 to 2026-05-09."}, {"sent_id": 10, "text": "Marble-8B was the model selected by Project Sonata on 2026-04-15 for the error analysis track."}], "source": "synthetic_rewritten", "title": "Review Packet 388"}
{"doc_id": "doc_000389", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron's Milestone D2 deadline change on 2026-06-04 replaced 2026-06-14 with 2026-06-24."}, {"sent_id": 3, "text": "The 2026-05-30 result for Project Sonata put Aster-7B on RiverBench-3 at 0.604 evidence F1, with Node Willow-05 handling the run."}, {"sent_id": 4, "text": "Project Aster received 1 GPU from Node Poplar-12 in the 2026-06-05 allocation log."}, {"sent_id": 5, "text": "The 2026-05-17 method entry for Project Meridian moved from chain verification to structured prompting."}, {"sent_id": 6, "text": "The staffing note for Project Meridian was dated 2026-06-30."}, {"sent_id": 7, "text": "It named Nadia Chen for the evaluation owner duties."}, {"sent_id": 8, "text": "The staffing note for Project Saffron was dated 2026-04-07."}, {"sent_id": 9, "text": "It named Priya Vale for the evaluation owner duties."}], "source": "synthetic_rewritten", "title": "Run Brief 389"}
{"doc_id": "doc_000390", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The staffing note for Project Meridian was dated 2026-06-07."}, {"sent_id": 3, "text": "It named Dr. Ravi Nadir for the data steward duties."}, {"sent_id": 4, "text": "The 2026-06-13 compute note links Node Fir-10 to Project Sonata with an allocation of 1 GPU."}, {"sent_id": 5, "text": "A compute allocation for Project Saffron was logged on 2026-06-06."}, {"sent_id": 6, "text": "Node Pine-07 supplied 4 GPUs."}, {"sent_id": 7, "text": "Project Saffron recorded a model choice for error analysis on 2026-06-03."}, {"sent_id": 8, "text": "The chosen model was Marble-3B."}, {"sent_id": 9, "text": "Using Node Elm-08, Nimbus-8B produced 0.708 macro F1 on TraceEval-2 for Project Nereid on 2026-05-30."}, {"sent_id": 10, "text": "Project Meridian had a method update on 2026-06-24."}, {"sent_id": 11, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 12, "text": "Using Node Fir-10, Vela-3B produced 0.655 latency efficiency score on LumenFacts-3 for Project Saffron on 2026-05-12."}, {"sent_id": 13, "text": "The 2026-04-27 compute note links Node Birch-04 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 14, "text": "In the 2026-04-28 staffing section, Project Meridian listed Dr. Ravi Reed as the evaluation owner."}], "source": "synthetic_rewritten", "title": "Lab Digest 390"}
{"doc_id": "doc_000391", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-04-15, Project Nereid retired rank fusion in favor of teacher distillation."}, {"sent_id": 3, "text": "Stable Chains Study 2 appeared in a 2026-05-22 review entry using chain verification; the note said it did not use a reward model."}, {"sent_id": 4, "text": "Project Aster ran Helix-7B on OrionBench-2 using Node Poplar-12 on 2026-05-18."}, {"sent_id": 5, "text": "The reported evidence F1 was 0.872."}, {"sent_id": 6, "text": "In the 2026-06-01 staffing section, Project Saffron listed Owen Marin as the data steward."}, {"sent_id": 7, "text": "For evidence retrieval, the 2026-06-07 note says Project Sonata went with Aster-3B."}, {"sent_id": 8, "text": "Project Meridian had a method update on 2026-05-17."}, {"sent_id": 9, "text": "The old approach was hard-negative mining, and the replacement was LoRA adaptation."}, {"sent_id": 10, "text": "Project Saffron ran Mira-3B on MemoTrace-2 using Node Sycamore-13 on 2026-04-20."}, {"sent_id": 11, "text": "The reported accuracy was 0.583."}], "source": "synthetic_rewritten", "title": "Project Update 391"}
{"doc_id": "doc_000392", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For error analysis, the 2026-06-18 note says Project Sonata went with Marble-7B."}, {"sent_id": 3, "text": "On 2026-04-12, Project Sonata retired dense retrieval in favor of rank fusion."}, {"sent_id": 4, "text": "Project Saffron had a method update on 2026-04-29."}, {"sent_id": 5, "text": "The old approach was teacher distillation, and the replacement was query rewriting."}, {"sent_id": 6, "text": "Project Sonata recorded a model choice for reranking on 2026-05-03."}, {"sent_id": 7, "text": "The chosen model was Vela-8B."}, {"sent_id": 8, "text": "The 2026-06-07 method entry for Project Nereid moved from LoRA adaptation to alias expansion."}, {"sent_id": 9, "text": "Project Anchor ran Helix-8B on CedarQA-3 using Node Poplar-12 on 2026-05-10."}, {"sent_id": 10, "text": "The reported latency efficiency score was 0.587."}, {"sent_id": 11, "text": "On 2026-06-26, Project Sonata shifted Milestone R1: 2026-07-05 became 2026-07-14."}, {"sent_id": 12, "text": "Using Node Birch-04, Kestrel-7B produced 0.732 macro F1 on OrionBench-3 for Project Nereid on 2026-06-15."}], "source": "synthetic_rewritten", "title": "Lab Digest 392"}
{"doc_id": "doc_000393", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For Project Anchor, the run using River-8B on Node Rowan-09 did not complete on 2026-04-10; the issue was a missing-index error."}, {"sent_id": 3, "text": "On 2026-04-18, Project Anchor retired query rewriting in favor of sentence pruning."}, {"sent_id": 4, "text": "In the 2026-06-08 staffing section, Project Anchor listed Dr. Arun Kwan as the data steward."}, {"sent_id": 5, "text": "The 2026-05-26 result for Project Anchor put Lumen-7B on MemoTrace-3 at 0.533 latency efficiency score, with Node Spruce-03 handling the run."}, {"sent_id": 6, "text": "Project Meridian logged a personnel update on 2026-04-21: Dr. Ravi Nadir took the data steward responsibilities."}, {"sent_id": 7, "text": "In the 2026-06-29 staffing section, Project Aster listed Anika Costa as the retrieval owner."}, {"sent_id": 8, "text": "The 2026-05-14 failure note says Marble-3B on Node Maple-01 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 9, "text": "Using Node Cedar-02, Orchid-7B produced 0.830 macro F1 on LabQA-3 for Project Aster on 2026-05-16."}, {"sent_id": 10, "text": "In the 2026-06-29 staffing section, Project Nereid listed Darian Grant as the lead."}], "source": "synthetic_rewritten", "title": "Project Update 393"}
{"doc_id": "doc_000394", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "For error analysis, the 2026-05-27 note says Project Aster went with Finch-7B."}, {"sent_id": 3, "text": "For claim classification, the 2026-05-15 note says Project Meridian went with Kestrel-3B."}, {"sent_id": 4, "text": "The 2026-05-22 failure note says Marble-7B on Node Maple-01 failed for Project Sonata because of an out-of-memory error."}, {"sent_id": 5, "text": "Milestone H1 for Project Anchor was revised on 2026-05-06."}, {"sent_id": 6, "text": "The schedule moved from 2026-05-20 to 2026-05-30."}, {"sent_id": 7, "text": "On 2026-06-24, Project Anchor shifted Milestone N1: 2026-07-16 became 2026-07-26."}, {"sent_id": 8, "text": "The 2026-04-17 failure note says Cedar-8B on Node Laurel-11 failed for Project Nereid because of an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Lab Digest 394"}
{"doc_id": "doc_000395", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-04-05 staffing section, Project Aster listed Felix Lind as the lead."}, {"sent_id": 3, "text": "Project Anchor's Milestone T1 deadline change on 2026-05-27 replaced 2026-06-06 with 2026-06-12."}, {"sent_id": 4, "text": "Project Aster had a method update on 2026-05-20."}, {"sent_id": 5, "text": "The old approach was query rewriting, and the replacement was sentence pruning."}, {"sent_id": 6, "text": "Aster-8B was the model selected by Project Saffron on 2026-05-31 for the evidence retrieval track."}, {"sent_id": 7, "text": "The 2026-06-04 failure note says Finch-8B on Node Juniper-06 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 8, "text": "The 2026-06-21 method entry for Project Sonata moved from chain verification to structured prompting."}, {"sent_id": 9, "text": "Helix-3B was the model selected by Project Aster on 2026-05-10 for the claim classification track."}, {"sent_id": 10, "text": "A compute allocation for Project Saffron was logged on 2026-06-04."}, {"sent_id": 11, "text": "Node Fir-10 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Project Update 395"}
{"doc_id": "doc_000396", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Meridian was logged on 2026-04-18."}, {"sent_id": 3, "text": "Node Hazel-14 supplied 2 GPUs."}, {"sent_id": 4, "text": "Milestone D1 for Project Nereid was revised on 2026-04-16."}, {"sent_id": 5, "text": "The schedule moved from 2026-05-03 to 2026-05-12."}, {"sent_id": 6, "text": "Project Aster logged a personnel update on 2026-04-26: Mina Shah took the lead responsibilities."}, {"sent_id": 7, "text": "Project Nereid's Milestone D1 deadline change on 2026-04-30 replaced 2026-05-17 with 2026-05-22."}, {"sent_id": 8, "text": "The staffing note for Project Meridian was dated 2026-04-12."}, {"sent_id": 9, "text": "It named Camila Quinn for the evaluation owner duties."}, {"sent_id": 10, "text": "The 2026-05-14 failure note says Nimbus-3B on Node Elm-08 failed for Project Nereid because of an out-of-memory error."}, {"sent_id": 11, "text": "On 2026-05-16, Project Aster retired late interaction in favor of BM25 retrieval."}, {"sent_id": 12, "text": "Using Node Sycamore-13, Mira-8B produced 0.688 macro F1 on TraceEval-3 for Project Sonata on 2026-06-16."}], "source": "synthetic_rewritten", "title": "Lab Digest 396"}
{"doc_id": "doc_000397", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-05-19 staffing section, Project Nereid listed Darian Hale as the retrieval owner."}, {"sent_id": 3, "text": "The 2026-05-18 result for Project Saffron put Vela-3B on LumenFacts-3 at 0.625 latency efficiency score, with Node Fir-10 handling the run."}, {"sent_id": 4, "text": "Project Sonata's Milestone F1 deadline change on 2026-05-04 replaced 2026-05-17 with 2026-05-22."}, {"sent_id": 5, "text": "For error analysis, the 2026-06-07 note says Project Aster went with Finch-8B."}, {"sent_id": 6, "text": "Project Anchor recorded a model choice for claim classification on 2026-05-01."}, {"sent_id": 7, "text": "The chosen model was Helix-7B."}, {"sent_id": 8, "text": "Project Nereid's Milestone P1 deadline change on 2026-04-07 replaced 2026-04-20 with 2026-04-29."}, {"sent_id": 9, "text": "Project Anchor had a method update on 2026-04-12."}, {"sent_id": 10, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}], "source": "synthetic_rewritten", "title": "Project Update 397"}
{"doc_id": "doc_000398", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For calibration, the 2026-05-22 note says Project Nereid went with Nimbus-7B."}, {"sent_id": 3, "text": "For calibration, the 2026-05-20 note says Project Meridian went with Nimbus-8B."}, {"sent_id": 4, "text": "The 2026-06-19 failure note says Helix-7B on Node Poplar-12 failed for Project Anchor because of a missing-index error."}, {"sent_id": 5, "text": "For evidence retrieval, the 2026-04-12 note says Project Nereid went with Atlas-3B."}, {"sent_id": 6, "text": "Project Saffron had a method update on 2026-05-31."}, {"sent_id": 7, "text": "The old approach was cross-encoder reranking, and the replacement was late interaction."}, {"sent_id": 8, "text": "Project Anchor logged a personnel update on 2026-04-13: Iris Stone took the data steward responsibilities."}, {"sent_id": 9, "text": "Using Node Juniper-06, Finch-8B produced 0.851 latency efficiency score on VestaLogs for Project Anchor on 2026-04-21."}], "source": "synthetic_rewritten", "title": "Lab Digest 398"}
{"doc_id": "doc_000399", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-24 method entry for Project Nereid moved from reward reranking to data mixing."}, {"sent_id": 3, "text": "Project Sonata recorded a model choice for error analysis on 2026-06-26."}, {"sent_id": 4, "text": "The chosen model was Marble-8B."}, {"sent_id": 5, "text": "Project Nereid ran Nimbus-3B on TraceEval-2 using Node Elm-08 on 2026-06-13."}, {"sent_id": 6, "text": "The reported macro F1 was 0.768."}, {"sent_id": 7, "text": "The staffing note for Project Meridian was dated 2026-06-02."}, {"sent_id": 8, "text": "It named Dr. Ravi Nadir for the data steward duties."}, {"sent_id": 9, "text": "Milestone V1 for Project Nereid was revised on 2026-05-26."}, {"sent_id": 10, "text": "The schedule moved from 2026-06-16 to 2026-06-25."}, {"sent_id": 11, "text": "Project Nereid had a failed run on 2026-06-18 involving Kestrel-7B and Node Birch-04."}, {"sent_id": 12, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 13, "text": "Aster-8B was the model selected by Project Sonata on 2026-05-27 for the evidence retrieval track."}], "source": "synthetic_rewritten", "title": "Project Update 399"}
{"doc_id": "doc_000400", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-27 result for Project Meridian put Cedar-8B on VestaLogs-2 at 0.711 latency efficiency score, with Node Laurel-11 handling the run."}, {"sent_id": 3, "text": "On 2026-05-07, Project Nereid shifted Milestone P1: 2026-05-20 became 2026-05-25."}, {"sent_id": 4, "text": "Project Anchor's Milestone T1 deadline change on 2026-06-03 replaced 2026-06-23 with 2026-06-27."}, {"sent_id": 5, "text": "A compute allocation for Project Sonata was logged on 2026-05-29."}, {"sent_id": 6, "text": "Node Pine-07 supplied 1 GPU."}, {"sent_id": 7, "text": "For claim classification, the 2026-06-28 note says Project Sonata went with Quartz-7B."}, {"sent_id": 8, "text": "On 2026-05-29, Project Aster shifted Milestone T1: 2026-06-13 became 2026-06-09."}], "source": "synthetic_rewritten", "title": "Review Packet 400"}
{"doc_id": "doc_000401", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-06-14, Project Anchor retired reward reranking in favor of data mixing."}, {"sent_id": 3, "text": "For error analysis, the 2026-06-17 note says Project Nereid went with Cedar-8B."}, {"sent_id": 4, "text": "The staffing note for Project Aster was dated 2026-06-08."}, {"sent_id": 5, "text": "It named Mina Adler for the retrieval owner duties."}, {"sent_id": 6, "text": "For reranking, the 2026-05-15 note says Project Aster went with Orchid-8B."}, {"sent_id": 7, "text": "For calibration, the 2026-06-23 note says Project Anchor went with Lumen-7B."}, {"sent_id": 8, "text": "Project Sonata had a failed run on 2026-05-29 involving Mira-8B and Node Sycamore-13."}, {"sent_id": 9, "text": "The recorded cause was an out-of-memory error."}], "source": "synthetic_rewritten", "title": "Run Brief 401"}
{"doc_id": "doc_000402", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-18 result for Project Anchor put Lumen-3B on MemoTrace-3 at 0.833 latency efficiency score, with Node Spruce-03 handling the run."}, {"sent_id": 3, "text": "The staffing note for Project Sonata was dated 2026-04-28."}, {"sent_id": 4, "text": "It named Sofia Frost for the lead duties."}, {"sent_id": 5, "text": "On 2026-05-21, Project Sonata retired document chunking in favor of metric smoothing."}, {"sent_id": 6, "text": "For calibration, the 2026-05-22 note says Project Saffron went with Mira-7B."}, {"sent_id": 7, "text": "A compute allocation for Project Sonata was logged on 2026-05-07."}, {"sent_id": 8, "text": "Node Pine-07 supplied 1 GPU."}, {"sent_id": 9, "text": "In the 2026-06-15 staffing section, Project Aster listed Dr. Noah Vale as the lead."}, {"sent_id": 10, "text": "Project Aster recorded a model choice for error analysis on 2026-04-23."}, {"sent_id": 11, "text": "The chosen model was Finch-7B."}], "source": "synthetic_rewritten", "title": "Review Packet 402"}
{"doc_id": "doc_000403", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Milestone B1 for Project Anchor was revised on 2026-04-30."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-26 to 2026-06-05."}, {"sent_id": 4, "text": "Project Aster had a method update on 2026-05-16."}, {"sent_id": 5, "text": "The old approach was evidence pooling, and the replacement was calibrated voting."}, {"sent_id": 6, "text": "Project Sonata logged a personnel update on 2026-05-26: Mara Quinn took the lead responsibilities."}, {"sent_id": 7, "text": "Project Meridian had a failed run on 2026-04-23 involving Cedar-7B and Node Laurel-11."}, {"sent_id": 8, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "Project Sonata logged a personnel update on 2026-04-28: Mara Lane took the retrieval owner responsibilities."}, {"sent_id": 10, "text": "On 2026-05-06, Project Anchor retired teacher distillation in favor of query rewriting."}, {"sent_id": 11, "text": "Project Meridian ran Nova-8B on LumenFacts-2 using Node Aspen-01 on 2026-05-09."}, {"sent_id": 12, "text": "The reported latency efficiency score was 0.735."}], "source": "synthetic_rewritten", "title": "Run Brief 403"}
{"doc_id": "doc_000404", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Milestone Z1 for Project Aster was revised on 2026-04-23."}, {"sent_id": 3, "text": "The schedule moved from 2026-05-16 to 2026-05-14."}, {"sent_id": 4, "text": "On 2026-06-11, Project Meridian shifted Milestone D1: 2026-07-03 became 2026-07-09."}, {"sent_id": 5, "text": "The 2026-06-21 method entry for Project Meridian moved from sentence pruning to cross-encoder reranking."}, {"sent_id": 6, "text": "On 2026-04-17, Project Anchor shifted Milestone B1: 2026-05-03 became 2026-05-11."}, {"sent_id": 7, "text": "The 2026-05-03 method entry for Project Meridian moved from rank fusion to teacher distillation."}, {"sent_id": 8, "text": "Project Sonata received 1 GPU from Node Maple-01 in the 2026-06-12 allocation log."}], "source": "synthetic_rewritten", "title": "Review Packet 404"}
{"doc_id": "doc_000405", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Nereid ran Atlas-8B on RiverBench-2 using Node Hazel-14 on 2026-04-13."}, {"sent_id": 3, "text": "The reported evidence F1 was 0.774."}, {"sent_id": 4, "text": "Project Saffron recorded a model choice for reranking on 2026-06-03."}, {"sent_id": 5, "text": "The chosen model was Vela-3B."}, {"sent_id": 6, "text": "Project Saffron logged a personnel update on 2026-06-29: Priya Vale took the evaluation owner responsibilities."}, {"sent_id": 7, "text": "Project Saffron had a failed run on 2026-05-14 involving Vela-7B and Node Fir-10."}, {"sent_id": 8, "text": "The recorded cause was a missing-index error."}, {"sent_id": 9, "text": "Project Anchor had a method update on 2026-05-07."}, {"sent_id": 10, "text": "The old approach was BM25 retrieval, and the replacement was evidence pooling."}, {"sent_id": 11, "text": "The 2026-04-18 method entry for Project Aster moved from hard-negative mining to LoRA adaptation."}, {"sent_id": 12, "text": "The 2026-04-01 method entry for Project Anchor moved from BM25 retrieval to evidence pooling."}, {"sent_id": 13, "text": "Project Meridian logged a personnel update on 2026-04-07: Nadia Singh took the data steward responsibilities."}, {"sent_id": 14, "text": "Project Saffron received 2 GPUs from Node Maple-01 in the 2026-06-27 allocation log."}], "source": "synthetic_rewritten", "title": "Run Brief 405"}
{"doc_id": "doc_000406", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "For claim classification, the 2026-06-14 note says Project Nereid went with Kestrel-8B."}, {"sent_id": 3, "text": "Using Node Juniper-06, Finch-8B produced 0.821 latency efficiency score on VestaLogs for Project Anchor on 2026-04-27."}, {"sent_id": 4, "text": "On 2026-04-01, Project Anchor retired threshold search in favor of reward reranking."}, {"sent_id": 5, "text": "Project Sonata recorded a model choice for evidence retrieval on 2026-04-01."}, {"sent_id": 6, "text": "The chosen model was Aster-3B."}, {"sent_id": 7, "text": "Project Aster received 3 GPUs from Node Juniper-06 in the 2026-04-16 allocation log."}, {"sent_id": 8, "text": "Project Saffron recorded a model choice for error analysis on 2026-05-01."}, {"sent_id": 9, "text": "The chosen model was Marble-8B."}, {"sent_id": 10, "text": "For error analysis, the 2026-04-05 note says Project Saffron went with Marble-7B."}, {"sent_id": 11, "text": "Quiet Retriever Study 4 appeared in a 2026-06-24 review entry using sentence pruning; the note said it did not use a reward model."}, {"sent_id": 12, "text": "Using Node Elm-08, Nimbus-8B produced 0.573 latency efficiency score on MemoTrace for Project Meridian on 2026-05-05."}], "source": "synthetic_rewritten", "title": "Review Packet 406"}
{"doc_id": "doc_000407", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "A compute allocation for Project Meridian was logged on 2026-04-16."}, {"sent_id": 3, "text": "Node Elm-08 supplied 2 GPUs."}, {"sent_id": 4, "text": "The 2026-04-04 compute note links Node Poplar-12 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 5, "text": "Milestone N1 for Project Anchor was revised on 2026-05-08."}, {"sent_id": 6, "text": "The schedule moved from 2026-05-20 to 2026-05-24."}, {"sent_id": 7, "text": "Milestone V1 for Project Meridian was revised on 2026-05-14."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-30 to 2026-06-03."}, {"sent_id": 9, "text": "Project Sonata logged a personnel update on 2026-06-16: Sofia Nadir took the retrieval owner responsibilities."}, {"sent_id": 10, "text": "Project Meridian's Milestone B2 deadline change on 2026-04-09 replaced 2026-04-23 with 2026-05-03."}, {"sent_id": 11, "text": "For Project Saffron, the run using Vela-3B on Node Fir-10 did not complete on 2026-06-27; the issue was a checkpoint-mismatch error."}, {"sent_id": 12, "text": "Project Meridian received 2 GPUs from Node Laurel-11 in the 2026-05-09 allocation log."}], "source": "synthetic_rewritten", "title": "Run Brief 407"}
{"doc_id": "doc_000408", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Aster ran Lumen-8B on TraceEval using Node Spruce-03 on 2026-05-24."}, {"sent_id": 3, "text": "The reported macro F1 was 0.608."}, {"sent_id": 4, "text": "Milestone H1 for Project Anchor was revised on 2026-05-27."}, {"sent_id": 5, "text": "The schedule moved from 2026-06-20 to 2026-06-24."}, {"sent_id": 6, "text": "Project Meridian logged a personnel update on 2026-04-27: Elian Shah took the data steward responsibilities."}, {"sent_id": 7, "text": "For calibration, the 2026-06-07 note says Project Meridian went with Nimbus-8B."}, {"sent_id": 8, "text": "Using Node Aspen-01, Nova-7B produced 0.765 accuracy on LumenFacts-2 for Project Meridian on 2026-05-16."}, {"sent_id": 9, "text": "For evidence retrieval, the 2026-05-22 note says Project Anchor went with River-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 408"}
{"doc_id": "doc_000409", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian received 2 GPUs from Node Aspen-01 in the 2026-04-04 allocation log."}, {"sent_id": 3, "text": "Project Saffron's Milestone L1 deadline change on 2026-05-12 replaced 2026-06-07 with 2026-06-13."}, {"sent_id": 4, "text": "On 2026-05-08, Project Aster shifted Milestone H1: 2026-05-27 became 2026-05-24."}, {"sent_id": 5, "text": "Project Sonata recorded a model choice for claim classification on 2026-06-10."}, {"sent_id": 6, "text": "The chosen model was Quartz-7B."}, {"sent_id": 7, "text": "A compute allocation for Project Anchor was logged on 2026-05-02."}, {"sent_id": 8, "text": "Node Cedar-02 supplied 2 GPUs."}, {"sent_id": 9, "text": "Project Sonata recorded a model choice for calibration on 2026-05-06."}, {"sent_id": 10, "text": "The chosen model was Mira-7B."}, {"sent_id": 11, "text": "Project Meridian ran Atlas-3B on SignalSet-3 using Node Hazel-14 on 2026-05-11."}, {"sent_id": 12, "text": "The reported latency efficiency score was 0.699."}, {"sent_id": 13, "text": "Helix-3B was the model selected by Project Aster on 2026-05-28 for the claim classification track."}], "source": "synthetic_rewritten", "title": "Run Brief 409"}
{"doc_id": "doc_000410", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "On 2026-06-03, Project Saffron retired chain verification in favor of structured prompting."}, {"sent_id": 3, "text": "The 2026-06-14 method entry for Project Aster moved from contrastive tuning to threshold search."}, {"sent_id": 4, "text": "On 2026-04-03, the Quiet Retriever Study 3 review reported metric smoothing and said it did not use a reward model."}, {"sent_id": 5, "text": "A compute allocation for Project Sonata was logged on 2026-05-09."}, {"sent_id": 6, "text": "Node Maple-01 supplied 3 GPUs."}, {"sent_id": 7, "text": "Project Anchor logged a personnel update on 2026-05-19: Selene Rios took the data steward responsibilities."}, {"sent_id": 8, "text": "Project Sonata recorded a model choice for claim classification on 2026-04-12."}, {"sent_id": 9, "text": "The chosen model was Quartz-8B."}], "source": "synthetic_rewritten", "title": "Lab Digest 410"}
{"doc_id": "doc_000411", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "On 2026-06-21, Project Meridian retired document chunking in favor of metric smoothing."}, {"sent_id": 3, "text": "A compute allocation for Project Nereid was logged on 2026-05-07."}, {"sent_id": 4, "text": "Node Laurel-11 supplied 3 GPUs."}, {"sent_id": 5, "text": "A compute allocation for Project Anchor was logged on 2026-04-03."}, {"sent_id": 6, "text": "Node Cedar-02 supplied 4 GPUs."}, {"sent_id": 7, "text": "Using Node Birch-04, Kestrel-7B produced 0.762 evidence F1 on OrionBench-3 for Project Nereid on 2026-04-18."}, {"sent_id": 8, "text": "Atlas-7B was the model selected by Project Meridian on 2026-06-16 for the evidence retrieval track."}, {"sent_id": 9, "text": "Using Node Maple-01, Marble-7B produced 0.646 macro F1 on NereidNotes for Project Sonata on 2026-06-27."}, {"sent_id": 10, "text": "The 2026-06-20 result for Project Sonata put Mira-7B on TraceEval-3 at 0.538 macro F1, with Node Sycamore-13 handling the run."}], "source": "synthetic_rewritten", "title": "Project Update 411"}
{"doc_id": "doc_000412", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Meridian's Milestone B2 deadline change on 2026-06-03 replaced 2026-06-27 with 2026-07-05."}, {"sent_id": 3, "text": "For error analysis, the 2026-06-19 note says Project Sonata went with Marble-3B."}, {"sent_id": 4, "text": "Project Anchor ran Lumen-7B on MemoTrace-3 using Node Spruce-03 on 2026-04-05."}, {"sent_id": 5, "text": "The reported accuracy was 0.593."}, {"sent_id": 6, "text": "For calibration, the 2026-06-03 note says Project Aster went with Lumen-7B."}, {"sent_id": 7, "text": "Project Anchor recorded a model choice for claim classification on 2026-04-26."}, {"sent_id": 8, "text": "The chosen model was Helix-7B."}, {"sent_id": 9, "text": "The 2026-05-19 result for Project Nereid put Nimbus-7B on TraceEval-2 at 0.828 evidence F1, with Node Elm-08 handling the run."}, {"sent_id": 10, "text": "A compute allocation for Project Nereid was logged on 2026-05-28."}, {"sent_id": 11, "text": "Node Elm-08 supplied 3 GPUs."}], "source": "synthetic_rewritten", "title": "Lab Digest 412"}
{"doc_id": "doc_000413", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron's Milestone L1 deadline change on 2026-06-19 replaced 2026-07-15 with 2026-07-25."}, {"sent_id": 3, "text": "On 2026-05-06, Project Sonata retired hard-negative mining in favor of LoRA adaptation."}, {"sent_id": 4, "text": "Mira-8B was the model selected by Project Saffron on 2026-05-15 for the calibration track."}, {"sent_id": 5, "text": "In the 2026-06-02 staffing section, Project Aster listed Anika Costa as the retrieval owner."}, {"sent_id": 6, "text": "On 2026-04-25, Project Aster retired cross-encoder reranking in favor of late interaction."}, {"sent_id": 7, "text": "Project Saffron ran Mira-7B on MemoTrace-2 using Node Sycamore-13 on 2026-05-04."}, {"sent_id": 8, "text": "The reported accuracy was 0.643."}], "source": "synthetic_rewritten", "title": "Project Update 413"}
{"doc_id": "doc_000414", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The 2026-05-30 compute note links Node Poplar-12 to Project Anchor with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "For evidence retrieval, the 2026-06-14 note says Project Saffron went with Aster-3B."}, {"sent_id": 4, "text": "Using Node Spruce-03, Lumen-8B produced 0.548 evidence F1 on TraceEval for Project Aster on 2026-04-14."}, {"sent_id": 5, "text": "Project Meridian had a method update on 2026-04-08."}, {"sent_id": 6, "text": "The old approach was dense retrieval, and the replacement was rank fusion."}, {"sent_id": 7, "text": "Project Saffron ran Quartz-3B on CedarQA-2 using Node Pine-07 on 2026-05-25."}, {"sent_id": 8, "text": "The reported latency efficiency score was 0.577."}, {"sent_id": 9, "text": "The staffing note for Project Sonata was dated 2026-05-03."}, {"sent_id": 10, "text": "It named Samir Ames for the lead duties."}, {"sent_id": 11, "text": "Milestone F1 for Project Saffron was revised on 2026-04-01."}, {"sent_id": 12, "text": "The schedule moved from 2026-04-09 to 2026-04-13."}, {"sent_id": 13, "text": "Project Saffron received 4 GPUs from Node Fir-10 in the 2026-05-02 allocation log."}, {"sent_id": 14, "text": "Kestrel-8B was the model selected by Project Nereid on 2026-04-23 for the claim classification track."}], "source": "synthetic_rewritten", "title": "Lab Digest 414"}
{"doc_id": "doc_000415", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-05-15, the Stable Chains Study 1 review reported alias expansion and said it did not use a reward model."}, {"sent_id": 3, "text": "For evidence retrieval, the 2026-06-05 note says Project Sonata went with Aster-7B."}, {"sent_id": 4, "text": "Project Saffron logged a personnel update on 2026-06-16: Priya Moss took the data steward responsibilities."}, {"sent_id": 5, "text": "Nested Verifier Study 1 appeared in a 2026-05-18 review entry using calibrated voting; the note said it did not use a reward model."}, {"sent_id": 6, "text": "For Project Anchor, the run using Lumen-3B on Node Spruce-03 did not complete on 2026-05-15; the issue was a checkpoint-mismatch error."}, {"sent_id": 7, "text": "Project Meridian received 4 GPUs from Node Elm-08 in the 2026-05-01 allocation log."}, {"sent_id": 8, "text": "Project Meridian's Milestone J1 deadline change on 2026-06-22 replaced 2026-07-02 with 2026-07-12."}], "source": "synthetic_rewritten", "title": "Project Update 415"}
{"doc_id": "doc_000416", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Anchor logged a personnel update on 2026-06-15: Dr. Arun Bauer took the evaluation owner responsibilities."}, {"sent_id": 3, "text": "A compute allocation for Project Saffron was logged on 2026-05-16."}, {"sent_id": 4, "text": "Node Fir-10 supplied 4 GPUs."}, {"sent_id": 5, "text": "Project Saffron had a method update on 2026-05-24."}, {"sent_id": 6, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 7, "text": "In the 2026-05-12 staffing section, Project Aster listed Anika Sato as the lead."}, {"sent_id": 8, "text": "The 2026-04-23 compute note links Node Poplar-12 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 9, "text": "Project Aster had a failed run on 2026-05-23 involving Lumen-3B and Node Spruce-03."}, {"sent_id": 10, "text": "The recorded cause was an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Lab Digest 416"}
{"doc_id": "doc_000417", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Marble-7B was the model selected by Project Sonata on 2026-05-10 for the error analysis track."}, {"sent_id": 3, "text": "In the 2026-05-24 staffing section, Project Anchor listed Selene Rios as the data steward."}, {"sent_id": 4, "text": "Using Node Spruce-03, Lumen-8B produced 0.848 macro F1 on TraceEval for Project Aster on 2026-06-23."}, {"sent_id": 5, "text": "Project Meridian recorded a model choice for reranking on 2026-04-29."}, {"sent_id": 6, "text": "The chosen model was Nova-7B."}, {"sent_id": 7, "text": "Project Aster's Milestone N1 deadline change on 2026-06-26 replaced 2026-07-03 with 2026-07-02."}, {"sent_id": 8, "text": "Project Aster recorded a model choice for evidence retrieval on 2026-06-24."}, {"sent_id": 9, "text": "The chosen model was River-8B."}, {"sent_id": 10, "text": "Project Saffron logged a personnel update on 2026-05-11: Priya Moss took the data steward responsibilities."}, {"sent_id": 11, "text": "For Project Nereid, the run using Nova-3B on Node Aspen-01 did not complete on 2026-04-04; the issue was an unstable-validation-loss error."}, {"sent_id": 12, "text": "Project Saffron's Milestone L1 deadline change on 2026-04-02 replaced 2026-04-18 with 2026-04-22."}], "source": "synthetic_rewritten", "title": "Project Update 417"}
{"doc_id": "doc_000418", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "In the 2026-06-08 staffing section, Project Sonata listed Sofia Nadir as the retrieval owner."}, {"sent_id": 3, "text": "On 2026-05-01, Project Anchor shifted Milestone B1: 2026-05-17 became 2026-05-21."}, {"sent_id": 4, "text": "The staffing note for Project Nereid was dated 2026-04-13."}, {"sent_id": 5, "text": "It named Darian Grant for the lead duties."}, {"sent_id": 6, "text": "Using Node Aspen-01, Nova-3B produced 0.570 macro F1 on LabQA for Project Nereid on 2026-04-07."}, {"sent_id": 7, "text": "A compute allocation for Project Nereid was logged on 2026-05-29."}, {"sent_id": 8, "text": "Node Elm-08 supplied 1 GPU."}, {"sent_id": 9, "text": "Project Anchor's Milestone N1 deadline change on 2026-05-21 replaced 2026-06-12 with 2026-06-18."}, {"sent_id": 10, "text": "Milestone R1 for Project Sonata was revised on 2026-04-09."}, {"sent_id": 11, "text": "The schedule moved from 2026-04-28 to 2026-04-25."}, {"sent_id": 12, "text": "The staffing note for Project Saffron was dated 2026-05-31."}, {"sent_id": 13, "text": "It named Owen Marin for the data steward duties."}, {"sent_id": 14, "text": "The 2026-06-27 result for Project Sonata put Mira-3B on TraceEval-3 at 0.568 evidence F1, with Node Sycamore-13 handling the run."}], "source": "synthetic_rewritten", "title": "Lab Digest 418"}
{"doc_id": "doc_000419", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Saffron had a method update on 2026-06-24."}, {"sent_id": 3, "text": "The old approach was rank fusion, and the replacement was teacher distillation."}, {"sent_id": 4, "text": "For Project Nereid, the run using Kestrel-7B on Node Birch-04 did not complete on 2026-05-28; the issue was an out-of-memory error."}, {"sent_id": 5, "text": "For Project Meridian, the run using Atlas-8B on Node Hazel-14 did not complete on 2026-04-30; the issue was a checkpoint-mismatch error."}, {"sent_id": 6, "text": "On 2026-04-06, Project Nereid shifted Milestone P1: 2026-04-29 became 2026-04-27."}, {"sent_id": 7, "text": "Using Node Juniper-06, Finch-8B produced 0.716 macro F1 on NereidNotes-2 for Project Aster on 2026-05-05."}, {"sent_id": 8, "text": "On 2026-06-12, Project Saffron shifted Milestone D2: 2026-06-22 became 2026-07-02."}], "source": "synthetic_rewritten", "title": "Project Update 419"}
{"doc_id": "doc_000420", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "A compute allocation for Project Meridian was logged on 2026-05-15."}, {"sent_id": 3, "text": "Node Birch-04 supplied 4 GPUs."}, {"sent_id": 4, "text": "The 2026-06-05 compute note links Node Laurel-11 to Project Nereid with an allocation of 1 GPU."}, {"sent_id": 5, "text": "The staffing note for Project Anchor was dated 2026-04-07."}, {"sent_id": 6, "text": "It named Selene Rios for the data steward duties."}, {"sent_id": 7, "text": "The 2026-06-10 method entry for Project Anchor moved from structured prompting to QLoRA adaptation."}, {"sent_id": 8, "text": "Milestone Z1 for Project Anchor was revised on 2026-06-11."}, {"sent_id": 9, "text": "The schedule moved from 2026-06-29 to 2026-07-09."}, {"sent_id": 10, "text": "On 2026-06-17, Project Anchor shifted Milestone B1: 2026-07-13 became 2026-07-23."}, {"sent_id": 11, "text": "Project Sonata's Milestone F1 deadline change on 2026-06-18 replaced 2026-07-11 with 2026-07-09."}, {"sent_id": 12, "text": "Project Meridian's Milestone P1 deadline change on 2026-05-19 replaced 2026-06-06 with 2026-06-16."}], "source": "synthetic_rewritten", "title": "Run Brief 420"}
{"doc_id": "doc_000421", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "In the 2026-06-09 staffing section, Project Saffron listed Owen Marin as the data steward."}, {"sent_id": 3, "text": "Aster-7B was the model selected by Project Saffron on 2026-06-07 for the evidence retrieval track."}, {"sent_id": 4, "text": "A compute allocation for Project Nereid was logged on 2026-05-15."}, {"sent_id": 5, "text": "Node Hazel-14 supplied 1 GPU."}, {"sent_id": 6, "text": "For evidence retrieval, the 2026-04-03 note says Project Saffron went with Aster-8B."}, {"sent_id": 7, "text": "Milestone P1 for Project Meridian was revised on 2026-06-25."}, {"sent_id": 8, "text": "The schedule moved from 2026-07-03 to 2026-07-07."}, {"sent_id": 9, "text": "Project Nereid's Milestone J1 deadline change on 2026-04-02 replaced 2026-04-27 with 2026-05-06."}, {"sent_id": 10, "text": "For claim classification, the 2026-05-06 note says Project Saffron went with Quartz-7B."}], "source": "synthetic_rewritten", "title": "Review Packet 421"}
{"doc_id": "doc_000422", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Nova-3B was the model selected by Project Meridian on 2026-04-23 for the reranking track."}, {"sent_id": 3, "text": "The 2026-06-13 compute note links Node Juniper-06 to Project Aster with an allocation of 3 GPUs."}, {"sent_id": 4, "text": "A compute allocation for Project Anchor was logged on 2026-04-17."}, {"sent_id": 5, "text": "Node Poplar-12 supplied 2 GPUs."}, {"sent_id": 6, "text": "Using Node Laurel-11, Cedar-8B produced 0.771 latency efficiency score on VestaLogs-2 for Project Meridian on 2026-06-15."}, {"sent_id": 7, "text": "Project Anchor recorded a model choice for reranking on 2026-05-13."}, {"sent_id": 8, "text": "The chosen model was Orchid-3B."}, {"sent_id": 9, "text": "On 2026-05-28, Project Saffron shifted Milestone D2: 2026-06-17 became 2026-06-21."}, {"sent_id": 10, "text": "For Project Meridian, the run using Nova-3B on Node Aspen-01 did not complete on 2026-06-26; the issue was a checkpoint-mismatch error."}, {"sent_id": 11, "text": "On 2026-06-28, Project Nereid retired QLoRA adaptation in favor of hybrid retrieval."}, {"sent_id": 12, "text": "The 2026-05-07 failure note says Marble-7B on Node Maple-01 failed for Project Sonata because of an unstable-validation-loss error."}], "source": "synthetic_rewritten", "title": "Run Brief 422"}
{"doc_id": "doc_000423", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Meridian had a method update on 2026-06-21."}, {"sent_id": 3, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 4, "text": "On 2026-04-02, Project Saffron retired alias expansion in favor of dense retrieval."}, {"sent_id": 5, "text": "Project Aster logged a personnel update on 2026-06-01: Talia Reed took the retrieval owner responsibilities."}, {"sent_id": 6, "text": "The 2026-06-02 result for Project Sonata put Quartz-3B on OrionBench at 0.862 macro F1, with Node Pine-07 handling the run."}, {"sent_id": 7, "text": "The 2026-05-30 compute note links Node Laurel-11 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 8, "text": "Project Saffron received 2 GPUs from Node Sycamore-13 in the 2026-06-10 allocation log."}, {"sent_id": 9, "text": "The 2026-04-28 result for Project Aster put Orchid-7B on LabQA-3 at 0.530 macro F1, with Node Cedar-02 handling the run."}, {"sent_id": 10, "text": "Milestone J1 for Project Nereid was revised on 2026-06-11."}, {"sent_id": 11, "text": "The schedule moved from 2026-06-26 to 2026-06-22."}], "source": "synthetic_rewritten", "title": "Review Packet 423"}
{"doc_id": "doc_000424", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Nereid ran Kestrel-3B on OrionBench-3 using Node Birch-04 on 2026-06-22."}, {"sent_id": 3, "text": "The reported evidence F1 was 0.762."}, {"sent_id": 4, "text": "The 2026-04-27 result for Project Anchor put River-3B on SignalSet-2 at 0.899 accuracy, with Node Rowan-09 handling the run."}, {"sent_id": 5, "text": "In the 2026-06-16 staffing section, Project Aster listed Mina Adler as the retrieval owner."}, {"sent_id": 6, "text": "Using Node Fir-10, Vela-7B produced 0.580 evidence F1 on LabQA-2 for Project Sonata on 2026-04-18."}, {"sent_id": 7, "text": "The 2026-04-04 compute note links Node Fir-10 to Project Saffron with an allocation of 4 GPUs."}, {"sent_id": 8, "text": "On 2026-06-18, Project Aster shifted Milestone N1: 2026-06-25 became 2026-06-24."}, {"sent_id": 9, "text": "The 2026-04-17 failure note says Lumen-8B on Node Spruce-03 failed for Project Aster because of an out-of-memory error."}, {"sent_id": 10, "text": "For Project Anchor, the run using Lumen-7B on Node Spruce-03 did not complete on 2026-04-11; the issue was a checkpoint-mismatch error."}, {"sent_id": 11, "text": "Using Node Poplar-12, Helix-3B produced 0.707 latency efficiency score on CedarQA-3 for Project Anchor on 2026-05-12."}], "source": "synthetic_rewritten", "title": "Run Brief 424"}
{"doc_id": "doc_000425", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The staffing note for Project Sonata was dated 2026-06-02."}, {"sent_id": 3, "text": "It named Sofia Nadir for the retrieval owner duties."}, {"sent_id": 4, "text": "Project Nereid logged a personnel update on 2026-06-16: Nora Sol took the retrieval owner responsibilities."}, {"sent_id": 5, "text": "Project Sonata received 1 GPU from Node Sycamore-13 in the 2026-05-23 allocation log."}, {"sent_id": 6, "text": "Cedar-7B was the model selected by Project Meridian on 2026-04-29 for the error analysis track."}, {"sent_id": 7, "text": "The 2026-04-23 failure note says Cedar-8B on Node Laurel-11 failed for Project Nereid because of an out-of-memory error."}, {"sent_id": 8, "text": "Nova-7B was the model selected by Project Nereid on 2026-04-02 for the reranking track."}, {"sent_id": 9, "text": "In the 2026-04-20 staffing section, Project Nereid listed Dr. Julian Gray as the lead."}, {"sent_id": 10, "text": "On 2026-06-14, Project Saffron retired rank fusion in favor of teacher distillation."}, {"sent_id": 11, "text": "In the 2026-05-26 staffing section, Project Aster listed Felix Brooks as the retrieval owner."}], "source": "synthetic_rewritten", "title": "Review Packet 425"}
{"doc_id": "doc_000426", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Project Nereid logged a personnel update on 2026-06-23: Nora Bauer took the lead responsibilities."}, {"sent_id": 3, "text": "Lumen-8B was the model selected by Project Anchor on 2026-06-18 for the calibration track."}, {"sent_id": 4, "text": "The 2026-04-04 method entry for Project Aster moved from threshold search to reward reranking."}, {"sent_id": 5, "text": "The 2026-05-18 compute note links Node Maple-01 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 6, "text": "The 2026-04-13 compute note links Node Laurel-11 to Project Meridian with an allocation of 2 GPUs."}, {"sent_id": 7, "text": "On 2026-04-03, Project Anchor shifted Milestone T1: 2026-04-13 became 2026-04-23."}, {"sent_id": 8, "text": "Milestone D2 for Project Saffron was revised on 2026-04-21."}, {"sent_id": 9, "text": "The schedule moved from 2026-05-01 to 2026-05-11."}, {"sent_id": 10, "text": "For reranking, the 2026-04-17 note says Project Anchor went with Orchid-7B."}], "source": "synthetic_rewritten", "title": "Run Brief 426"}
{"doc_id": "doc_000427", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The staffing note for Project Saffron was dated 2026-06-15."}, {"sent_id": 3, "text": "It named Priya Vale for the evaluation owner duties."}, {"sent_id": 4, "text": "On 2026-05-03, Project Saffron retired hard-negative mining in favor of LoRA adaptation."}, {"sent_id": 5, "text": "The 2026-05-13 compute note links Node Poplar-12 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 6, "text": "The 2026-05-05 result for Project Aster put Helix-8B on OrionBench-2 at 0.872 macro F1, with Node Poplar-12 handling the run."}, {"sent_id": 7, "text": "Project Aster ran Orchid-3B on LabQA-3 using Node Cedar-02 on 2026-04-14."}, {"sent_id": 8, "text": "The reported macro F1 was 0.860."}, {"sent_id": 9, "text": "Project Aster logged a personnel update on 2026-05-12: Felix Brooks took the retrieval owner responsibilities."}, {"sent_id": 10, "text": "The 2026-05-02 failure note says Nimbus-8B on Node Elm-08 failed for Project Meridian because of a checkpoint-mismatch error."}], "source": "synthetic_rewritten", "title": "Review Packet 427"}
{"doc_id": "doc_000428", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-06-08 result for Project Nereid put Kestrel-8B on OrionBench-3 at 0.702 evidence F1, with Node Birch-04 handling the run."}, {"sent_id": 3, "text": "Project Anchor received 4 GPUs from Node Cedar-02 in the 2026-04-27 allocation log."}, {"sent_id": 4, "text": "The staffing note for Project Sonata was dated 2026-05-25."}, {"sent_id": 5, "text": "It named Dr. Jonas Rios for the lead duties."}, {"sent_id": 6, "text": "Project Anchor ran Helix-7B on CedarQA-3 using Node Poplar-12 on 2026-06-16."}, {"sent_id": 7, "text": "The reported accuracy was 0.857."}, {"sent_id": 8, "text": "The 2026-04-01 method entry for Project Meridian moved from confidence calibration to chain verification."}, {"sent_id": 9, "text": "A compute allocation for Project Saffron was logged on 2026-05-01."}, {"sent_id": 10, "text": "Node Maple-01 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Run Brief 428"}
{"doc_id": "doc_000429", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "In the 2026-04-20 staffing section, Project Saffron listed Dr. Theo Lind as the evaluation owner."}, {"sent_id": 3, "text": "For calibration, the 2026-06-19 note says Project Sonata went with Mira-8B."}, {"sent_id": 4, "text": "In the 2026-06-28 staffing section, Project Aster listed Dr. Noah Vale as the lead."}, {"sent_id": 5, "text": "Project Anchor logged a personnel update on 2026-05-25: Selene Rios took the data steward responsibilities."}, {"sent_id": 6, "text": "In the 2026-04-05 staffing section, Project Nereid listed Darian Hale as the retrieval owner."}, {"sent_id": 7, "text": "Project Saffron logged a personnel update on 2026-04-07: Dr. Theo Grant took the data steward responsibilities."}, {"sent_id": 8, "text": "Project Sonata recorded a model choice for calibration on 2026-05-20."}, {"sent_id": 9, "text": "The chosen model was Mira-8B."}], "source": "synthetic_rewritten", "title": "Review Packet 429"}
{"doc_id": "doc_000430", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Helix-3B was the model selected by Project Anchor on 2026-04-17 for the claim classification track."}, {"sent_id": 3, "text": "The 2026-04-10 failure note says Mira-7B on Node Sycamore-13 failed for Project Saffron because of a checkpoint-mismatch error."}, {"sent_id": 4, "text": "Project Saffron recorded a model choice for claim classification on 2026-06-18."}, {"sent_id": 5, "text": "The chosen model was Quartz-3B."}, {"sent_id": 6, "text": "River-3B was the model selected by Project Aster on 2026-06-25 for the evidence retrieval track."}, {"sent_id": 7, "text": "Project Nereid had a method update on 2026-04-30."}, {"sent_id": 8, "text": "The old approach was data mixing, and the replacement was contrastive tuning."}, {"sent_id": 9, "text": "Using Node Willow-05, Aster-3B produced 0.739 accuracy on SignalSet for Project Saffron on 2026-05-03."}], "source": "synthetic_rewritten", "title": "Project Update 430"}
{"doc_id": "doc_000431", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "Project Saffron's Milestone L1 deadline change on 2026-06-18 replaced 2026-07-04 with 2026-07-12."}, {"sent_id": 3, "text": "Project Aster recorded a model choice for calibration on 2026-05-07."}, {"sent_id": 4, "text": "The chosen model was Lumen-3B."}, {"sent_id": 5, "text": "Project Aster's Milestone N1 deadline change on 2026-05-27 replaced 2026-06-13 with 2026-06-18."}, {"sent_id": 6, "text": "Project Aster received 1 GPU from Node Cedar-02 in the 2026-06-12 allocation log."}, {"sent_id": 7, "text": "The 2026-05-30 result for Project Aster put Orchid-8B on LabQA-3 at 0.890 macro F1, with Node Cedar-02 handling the run."}, {"sent_id": 8, "text": "Using Node Fir-10, Vela-3B produced 0.820 evidence F1 on LabQA-2 for Project Sonata on 2026-05-05."}, {"sent_id": 9, "text": "Project Aster had a method update on 2026-06-24."}, {"sent_id": 10, "text": "The old approach was hybrid retrieval, and the replacement was document chunking."}, {"sent_id": 11, "text": "On 2026-04-05, Project Meridian retired LoRA adaptation in favor of alias expansion."}], "source": "synthetic_rewritten", "title": "Lab Digest 431"}
{"doc_id": "doc_000432", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "On 2026-05-08, Project Meridian shifted Milestone J1: 2026-05-18 became 2026-05-24."}, {"sent_id": 3, "text": "Vela-7B was the model selected by Project Sonata on 2026-06-21 for the reranking track."}, {"sent_id": 4, "text": "For Project Anchor, the run using Orchid-8B on Node Cedar-02 did not complete on 2026-06-19; the issue was a checkpoint-mismatch error."}, {"sent_id": 5, "text": "The 2026-05-10 method entry for Project Sonata moved from document chunking to metric smoothing."}, {"sent_id": 6, "text": "Project Aster received 1 GPU from Node Spruce-03 in the 2026-05-22 allocation log."}, {"sent_id": 7, "text": "Using Node Rowan-09, River-3B produced 0.704 macro F1 on RiverBench for Project Aster on 2026-04-14."}, {"sent_id": 8, "text": "On 2026-06-05, Project Aster shifted Milestone B1: 2026-06-16 became 2026-06-11."}, {"sent_id": 9, "text": "A compute allocation for Project Saffron was logged on 2026-04-04."}, {"sent_id": 10, "text": "Node Maple-01 supplied 4 GPUs."}], "source": "synthetic_rewritten", "title": "Project Update 432"}
{"doc_id": "doc_000433", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Elm-08, Nimbus-8B produced 0.678 evidence F1 on TraceEval-2 for Project Nereid on 2026-04-27."}, {"sent_id": 3, "text": "Orchid-8B was the model selected by Project Anchor on 2026-06-07 for the reranking track."}, {"sent_id": 4, "text": "Using Node Rowan-09, River-3B produced 0.539 accuracy on SignalSet-2 for Project Anchor on 2026-04-21."}, {"sent_id": 5, "text": "For error analysis, the 2026-05-13 note says Project Aster went with Finch-3B."}, {"sent_id": 6, "text": "On 2026-05-24, Project Saffron retired teacher distillation in favor of query rewriting."}, {"sent_id": 7, "text": "The 2026-04-30 method entry for Project Anchor moved from document chunking to metric smoothing."}], "source": "synthetic_rewritten", "title": "Lab Digest 433"}
{"doc_id": "doc_000434", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "The staffing note for Project Sonata was dated 2026-04-27."}, {"sent_id": 3, "text": "It named Sofia Nadir for the retrieval owner duties."}, {"sent_id": 4, "text": "Project Saffron recorded a model choice for evidence retrieval on 2026-06-21."}, {"sent_id": 5, "text": "The chosen model was Aster-8B."}, {"sent_id": 6, "text": "The staffing note for Project Saffron was dated 2026-05-26."}, {"sent_id": 7, "text": "It named Petra Gray for the evaluation owner duties."}, {"sent_id": 8, "text": "The 2026-06-15 result for Project Anchor put Orchid-3B on LumenFacts at 0.875 accuracy, with Node Cedar-02 handling the run."}, {"sent_id": 9, "text": "On 2026-05-10, Project Nereid retired threshold search in favor of reward reranking."}, {"sent_id": 10, "text": "The 2026-06-20 compute note links Node Juniper-06 to Project Aster with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Project Update 434"}
{"doc_id": "doc_000435", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-30 compute note links Node Maple-01 to Project Saffron with an allocation of 2 GPUs."}, {"sent_id": 3, "text": "The staffing note for Project Saffron was dated 2026-06-08."}, {"sent_id": 4, "text": "It named Petra Adler for the data steward duties."}, {"sent_id": 5, "text": "Kestrel-7B was the model selected by Project Nereid on 2026-04-03 for the claim classification track."}, {"sent_id": 6, "text": "Using Node Juniper-06, Finch-8B produced 0.866 macro F1 on NereidNotes-2 for Project Aster on 2026-06-22."}, {"sent_id": 7, "text": "On 2026-05-07, Project Saffron retired calibrated voting in favor of confidence calibration."}, {"sent_id": 8, "text": "For Project Anchor, the run using Orchid-7B on Node Cedar-02 did not complete on 2026-04-30; the issue was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "The staffing note for Project Meridian was dated 2026-06-16."}, {"sent_id": 10, "text": "It named Dr. Ravi Nadir for the data steward duties."}], "source": "synthetic_rewritten", "title": "Lab Digest 435"}
{"doc_id": "doc_000436", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Using Node Aspen-01, Nova-7B produced 0.720 evidence F1 on LabQA for Project Nereid on 2026-05-12."}, {"sent_id": 3, "text": "Project Anchor recorded a model choice for calibration on 2026-05-10."}, {"sent_id": 4, "text": "The chosen model was Lumen-8B."}, {"sent_id": 5, "text": "Project Nereid received 1 GPU from Node Elm-08 in the 2026-06-26 allocation log."}, {"sent_id": 6, "text": "Project Meridian logged a personnel update on 2026-04-28: Nadia Singh took the data steward responsibilities."}, {"sent_id": 7, "text": "Project Saffron's Milestone D2 deadline change on 2026-06-11 replaced 2026-07-01 with 2026-07-09."}, {"sent_id": 8, "text": "Project Nereid's Milestone D1 deadline change on 2026-06-24 replaced 2026-07-01 with 2026-06-30."}, {"sent_id": 9, "text": "Project Meridian received 4 GPUs from Node Hazel-14 in the 2026-04-11 allocation log."}, {"sent_id": 10, "text": "In the 2026-06-08 staffing section, Project Meridian listed Elian Ames as the evaluation owner."}, {"sent_id": 11, "text": "Project Saffron ran Aster-7B on SignalSet using Node Willow-05 on 2026-04-26."}, {"sent_id": 12, "text": "The reported latency efficiency score was 0.709."}], "source": "synthetic_rewritten", "title": "Project Update 436"}
{"doc_id": "doc_000437", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 9, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For evidence retrieval, the 2026-05-10 note says Project Nereid went with Atlas-3B."}, {"sent_id": 3, "text": "Project Meridian ran Kestrel-8B on CedarQA using Node Birch-04 on 2026-05-16."}, {"sent_id": 4, "text": "The reported latency efficiency score was 0.687."}, {"sent_id": 5, "text": "Milestone N1 for Project Anchor was revised on 2026-04-15."}, {"sent_id": 6, "text": "The schedule moved from 2026-04-27 to 2026-05-01."}, {"sent_id": 7, "text": "For Project Aster, the run using Finch-8B on Node Juniper-06 did not complete on 2026-05-02; the issue was an unstable-validation-loss error."}, {"sent_id": 8, "text": "Aster-3B was the model selected by Project Sonata on 2026-05-10 for the evidence retrieval track."}, {"sent_id": 9, "text": "For calibration, the 2026-04-03 note says Project Sonata went with Mira-3B."}, {"sent_id": 10, "text": "The 2026-06-24 compute note links Node Juniper-06 to Project Anchor with an allocation of 2 GPUs."}, {"sent_id": 11, "text": "Project Anchor ran Orchid-7B on LumenFacts using Node Cedar-02 on 2026-06-02."}, {"sent_id": 12, "text": "The reported latency efficiency score was 0.875."}, {"sent_id": 13, "text": "On 2026-04-06, Project Saffron shifted Milestone R1: 2026-04-20 became 2026-04-30."}], "source": "synthetic_rewritten", "title": "Lab Digest 437"}
{"doc_id": "doc_000438", "doc_type": "project_update", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "The 2026-05-29 compute note links Node Laurel-11 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 3, "text": "Project Nereid logged a personnel update on 2026-05-11: Dr. Julian Stone took the retrieval owner responsibilities."}, {"sent_id": 4, "text": "On 2026-04-05, Project Saffron retired dense retrieval in favor of rank fusion."}, {"sent_id": 5, "text": "A compute allocation for Project Nereid was logged on 2026-06-20."}, {"sent_id": 6, "text": "Node Hazel-14 supplied 1 GPU."}, {"sent_id": 7, "text": "The 2026-05-02 compute note links Node Aspen-01 to Project Nereid with an allocation of 3 GPUs."}, {"sent_id": 8, "text": "For Project Saffron, the run using Quartz-3B on Node Pine-07 did not complete on 2026-04-16; the issue was a checkpoint-mismatch error."}, {"sent_id": 9, "text": "In the 2026-04-20 staffing section, Project Aster listed Mina Adler as the retrieval owner."}], "source": "synthetic_rewritten", "title": "Project Update 438"}
{"doc_id": "doc_000439", "doc_type": "lab_digest", "metadata": {"fictional": true, "num_source_facts": 8, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "Nova-7B was the model selected by Project Nereid on 2026-05-24 for the reranking track."}, {"sent_id": 3, "text": "Using Node Hazel-14, Atlas-3B produced 0.744 macro F1 on RiverBench-2 for Project Nereid on 2026-04-06."}, {"sent_id": 4, "text": "Project Sonata had a method update on 2026-05-14."}, {"sent_id": 5, "text": "The old approach was query rewriting, and the replacement was sentence pruning."}, {"sent_id": 6, "text": "Project Saffron ran Quartz-7B on CedarQA-2 using Node Pine-07 on 2026-06-02."}, {"sent_id": 7, "text": "The reported latency efficiency score was 0.667."}, {"sent_id": 8, "text": "The 2026-06-28 method entry for Project Sonata moved from contrastive tuning to threshold search."}, {"sent_id": 9, "text": "The 2026-05-11 result for Project Saffron put Vela-7B on LumenFacts-3 at 0.595 accuracy, with Node Fir-10 handling the run."}, {"sent_id": 10, "text": "The 2026-06-24 method entry for Project Aster moved from cross-encoder reranking to late interaction."}, {"sent_id": 11, "text": "Project Sonata ran Quartz-7B on OrionBench using Node Pine-07 on 2026-05-26."}, {"sent_id": 12, "text": "The reported evidence F1 was 0.832."}], "source": "synthetic_rewritten", "title": "Lab Digest 439"}
{"doc_id": "doc_000440", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "A compute allocation for Project Meridian was logged on 2026-05-16."}, {"sent_id": 3, "text": "Node Laurel-11 supplied 4 GPUs."}, {"sent_id": 4, "text": "On 2026-05-28, Project Aster shifted Milestone B1: 2026-06-08 became 2026-06-03."}, {"sent_id": 5, "text": "Project Saffron ran Mira-7B on MemoTrace-2 using Node Sycamore-13 on 2026-05-16."}, {"sent_id": 6, "text": "The reported accuracy was 0.583."}, {"sent_id": 7, "text": "On 2026-05-05, Project Nereid shifted Milestone J1: 2026-05-30 became 2026-06-04."}, {"sent_id": 8, "text": "On 2026-05-10, Project Nereid retired chain verification in favor of structured prompting."}, {"sent_id": 9, "text": "On 2026-06-11, Project Anchor retired rank fusion in favor of teacher distillation."}], "source": "synthetic_rewritten", "title": "Review Packet 440"}
{"doc_id": "doc_000441", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "This note consolidates several project, model, compute, and review updates from the lab records."}, {"sent_id": 2, "text": "For calibration, the 2026-04-23 note says Project Meridian went with Nimbus-7B."}, {"sent_id": 3, "text": "Project Nereid received 1 GPU from Node Aspen-01 in the 2026-06-06 allocation log."}, {"sent_id": 4, "text": "Project Nereid ran Nova-8B on LabQA using Node Aspen-01 on 2026-04-04."}, {"sent_id": 5, "text": "The reported macro F1 was 0.780."}, {"sent_id": 6, "text": "Project Sonata had a failed run on 2026-04-02 involving Quartz-3B and Node Pine-07."}, {"sent_id": 7, "text": "The recorded cause was an out-of-memory error."}, {"sent_id": 8, "text": "Project Saffron logged a personnel update on 2026-04-07: Owen Torres took the evaluation owner responsibilities."}, {"sent_id": 9, "text": "In the 2026-05-18 staffing section, Project Meridian listed Elian Ames as the evaluation owner."}], "source": "synthetic_rewritten", "title": "Run Brief 441"}
{"doc_id": "doc_000442", "doc_type": "review_packet", "metadata": {"fictional": true, "num_source_facts": 7, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The digest below mixes staffing, schedule, model, and evaluation details from different workstreams."}, {"sent_id": 2, "text": "Project Saffron had a failed run on 2026-04-24 involving Mira-8B and Node Sycamore-13."}, {"sent_id": 3, "text": "The recorded cause was a checkpoint-mismatch error."}, {"sent_id": 4, "text": "The review note for Amber Ranking Study 2 was dated 2026-04-13."}, {"sent_id": 5, "text": "It described the method as hybrid retrieval and said it did not use a reward model."}, {"sent_id": 6, "text": "A compute allocation for Project Meridian was logged on 2026-06-12."}, {"sent_id": 7, "text": "Node Birch-04 supplied 4 GPUs."}, {"sent_id": 8, "text": "The 2026-06-21 method entry for Project Aster moved from document chunking to metric smoothing."}, {"sent_id": 9, "text": "On 2026-05-18, the Stable Chains Study 2 review reported BM25 retrieval and said it did not use a reward model."}, {"sent_id": 10, "text": "Project Anchor logged a personnel update on 2026-05-26: Selene Kim took the evaluation owner responsibilities."}, {"sent_id": 11, "text": "The 2026-06-27 compute note links Node Willow-05 to Project Sonata with an allocation of 1 GPU."}], "source": "synthetic_rewritten", "title": "Review Packet 442"}
{"doc_id": "doc_000443", "doc_type": "run_brief", "metadata": {"fictional": true, "num_source_facts": 6, "release_version": "rag-qa-v2-easy-rewritten"}, "sentences": [{"sent_id": 1, "text": "The record combines operational updates with experiment outcomes from the synthetic lab archive."}, {"sent_id": 2, "text": "The 2026-06-27 compute note links Node Aspen-01 to Project Meridian with an allocation of 4 GPUs."}, {"sent_id": 3, "text": "Project Aster had a failed run on 2026-06-19 involving Finch-8B and Node Juniper-06."}, {"sent_id": 4, "text": "The recorded cause was an unstable-validation-loss error."}, {"sent_id": 5, "text": "Milestone B2 for Project Nereid was revised on 2026-04-22."}, {"sent_id": 6, "text": "The schedule moved from 2026-05-11 to 2026-05-08."}, {"sent_id": 7, "text": "Milestone T1 for Project Aster was revised on 2026-04-14."}, {"sent_id": 8, "text": "The schedule moved from 2026-05-09 to 2026-05-14."}, {"sent_id": 9, "text": "For calibration, the 2026-05-31 note says Project Anchor went with Lumen-8B."}, {"sent_id": 10, "text": "The 2026-04-16 failure note says Kestrel-8B on Node Birch-04 failed for Project Meridian because of a missing-index error."}], "source": "synthetic_rewritten", "title": "Run Brief 443"}