diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index 475c5f8c5b162a9aab4c12474960d695cf787f2e..bc109b7807859d41ab841a38ece715741e6df3fb 100644 --- a/analysis/current/analysis-report-hybrid.json +++ b/analysis/current/analysis-report-hybrid.json @@ -1,8 +1,8 @@ { "schema_version": "1.0", "repo": "huggingface/transformers", - "snapshot_id": "20260721T000056Z", - "generated_at": "2026-07-21T00:24:00Z", + "snapshot_id": "20260721T060018Z", + "generated_at": "2026-07-21T06:15:09Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ @@ -1050,15 +1050,16 @@ ] }, { - "cluster_id": "cluster-46894-6", - "summary": "Cluster of 2 issues and 4 PRs centered on issue #46894.", + "cluster_id": "cluster-46894-7", + "summary": "Cluster of 3 issues and 4 PRs centered on issue #46894.", "status": "open", "confidence": 0.8, "canonical_issue_number": 46894, "canonical_pr_number": 46903, "issue_numbers": [ 46894, - 46897 + 46897, + 46901 ], "pr_numbers": [ 46895, @@ -1464,6 +1465,38 @@ } ] }, + { + "cluster_id": "cluster-43451-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #47321.", + "status": "open", + "confidence": 0.6, + "canonical_issue_number": 47321, + "canonical_pr_number": 47322, + "issue_numbers": [ + 47321 + ], + "pr_numbers": [ + 43451, + 47322 + ], + "evidence_types": [ + "closing_reference", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43451, + "right_pr_number": 47322, + "code_similarity": 0.001, + "size_similarity": 0.002, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.003, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, { "cluster_id": "cluster-41211-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #41211.", @@ -1564,159 +1597,216 @@ ] }, { - "cluster_id": "cluster-43665-6", - "summary": "Cluster of 1 issues and 5 PRs centered on issue #46869.", + "cluster_id": "cluster-46509-5", + "summary": "Cluster of 5 related pull requests linked by soft_similarity.", "status": "open", - "confidence": 0.8, - "canonical_issue_number": 46869, - "canonical_pr_number": 46886, - "issue_numbers": [ - 46869 - ], + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47143, + "issue_numbers": [], "pr_numbers": [ - 43665, - 44815, - 46380, - 46886, - 46922 + 46509, + 47143, + 47163, + 47301, + 47387 ], "evidence_types": [ - "closing_reference", - "shared_issue_target", "soft_similarity" ], "pr_comparisons": [ { - "left_pr_number": 43665, - "right_pr_number": 44815, - "code_similarity": 0.03, - "size_similarity": 0.133, + "left_pr_number": 46509, + "right_pr_number": 47143, + "code_similarity": 0.254, + "size_similarity": 0.793, + "file_overlap": 0.071, + "area_overlap": 0.143, + "patch_similarity": 0.161, + "shared_filenames": [ + "src/transformers/models/auto/auto_mappings.py", + "src/transformers/models/auto/modeling_auto.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/auto/auto_mappings.py", + "left_ranges": [ + [ + 412, + 418 + ], + [ + 1077, + 1083 + ], + [ + 1206, + 1212 + ] + ], + "right_ranges": [ + [ + 412, + 418 + ], + [ + 975, + 981 + ] + ] + }, + { + "filename": "src/transformers/models/auto/modeling_auto.py", + "left_ranges": [ + [ + 1097, + 1103 + ] + ], + "right_ranges": [ + [ + 348, + 354 + ], + [ + 1960, + 1966 + ] + ] + } + ] + }, + { + "left_pr_number": 46509, + "right_pr_number": 47163, + "code_similarity": 0.006, + "size_similarity": 0.012, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.022, + "patch_similarity": 0.024, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 43665, - "right_pr_number": 46380, - "code_similarity": 0.011, - "size_similarity": 0.048, + "left_pr_number": 46509, + "right_pr_number": 47301, + "code_similarity": 0.01, + "size_similarity": 0.013, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.006, + "patch_similarity": 0.047, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 43665, - "right_pr_number": 46886, - "code_similarity": 0.083, - "size_similarity": 0.414, + "left_pr_number": 46509, + "right_pr_number": 47387, + "code_similarity": 0.038, + "size_similarity": 0.119, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.0, + "patch_similarity": 0.095, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 43665, - "right_pr_number": 46922, - "code_similarity": 0.097, - "size_similarity": 0.483, + "left_pr_number": 47143, + "right_pr_number": 47163, + "code_similarity": 0.004, + "size_similarity": 0.009, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.0, + "patch_similarity": 0.013, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 44815, - "right_pr_number": 46380, - "code_similarity": 0.08, - "size_similarity": 0.363, + "left_pr_number": 47143, + "right_pr_number": 47301, + "code_similarity": 0.006, + "size_similarity": 0.01, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.048, + "patch_similarity": 0.026, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 44815, - "right_pr_number": 46886, - "code_similarity": 0.011, - "size_similarity": 0.055, + "left_pr_number": 47143, + "right_pr_number": 47387, + "code_similarity": 0.028, + "size_similarity": 0.095, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.0, + "patch_similarity": 0.06, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 44815, - "right_pr_number": 46922, - "code_similarity": 0.013, - "size_similarity": 0.064, + "left_pr_number": 47163, + "right_pr_number": 47301, + "code_similarity": 0.192, + "size_similarity": 0.933, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.004, + "patch_similarity": 0.033, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 46380, - "right_pr_number": 46886, - "code_similarity": 0.004, - "size_similarity": 0.02, + "left_pr_number": 47163, + "right_pr_number": 47387, + "code_similarity": 0.025, + "size_similarity": 0.098, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.0, + "patch_similarity": 0.033, "shared_filenames": [], "shared_file_areas": [] }, { - "left_pr_number": 46380, - "right_pr_number": 46922, - "code_similarity": 0.005, - "size_similarity": 0.023, + "left_pr_number": 47301, + "right_pr_number": 47387, + "code_similarity": 0.03, + "size_similarity": 0.105, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.0, + "patch_similarity": 0.059, "shared_filenames": [], "shared_file_areas": [] - }, + } + ] + }, + { + "cluster_id": "cluster-47332-4", + "summary": "Cluster of 2 issues and 2 PRs centered on issue #47333.", + "status": "open", + "confidence": 0.6, + "canonical_issue_number": 47333, + "canonical_pr_number": 47336, + "issue_numbers": [ + 47332, + 47333 + ], + "pr_numbers": [ + 47336, + 47383 + ], + "evidence_types": [ + "closing_reference", + "soft_similarity" + ], + "pr_comparisons": [ { - "left_pr_number": 46886, - "right_pr_number": 46922, - "code_similarity": 0.711, - "size_similarity": 0.857, - "file_overlap": 1.0, - "area_overlap": 0.5, - "patch_similarity": 0.429, - "shared_filenames": [ - "src/transformers/models/yolos/modeling_yolos.py" - ], - "shared_file_areas": [ - { - "filename": "src/transformers/models/yolos/modeling_yolos.py", - "left_ranges": [ - [ - 446, - 452 - ] - ], - "right_ranges": [ - [ - 434, - 440 - ], - [ - 446, - 452 - ] - ] - } - ] + "left_pr_number": 47336, + "right_pr_number": 47383, + "code_similarity": 0.019, + "size_similarity": 0.091, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.005, + "shared_filenames": [], + "shared_file_areas": [] } ] }, @@ -1820,253 +1910,48 @@ ] }, { - "cluster_id": "cluster-47077-4", - "summary": "Cluster of 4 related pull requests linked by soft_similarity.", + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 47077, - "issue_numbers": [], - "pr_numbers": [ - 47077, - 47210, - 47301, - 47304 - ], - "evidence_types": [ - "soft_similarity" + "confidence": 0.75, + "canonical_issue_number": 43240, + "canonical_pr_number": 43251, + "issue_numbers": [ + 43240 ], - "pr_comparisons": [ - { - "left_pr_number": 47077, - "right_pr_number": 47210, - "code_similarity": 0.208, - "size_similarity": 0.953, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.115, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 47077, - "right_pr_number": 47301, - "code_similarity": 0.03, - "size_similarity": 0.078, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.099, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 47077, - "right_pr_number": 47304, - "code_similarity": 0.06, - "size_similarity": 0.116, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.247, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 47210, - "right_pr_number": 47301, - "code_similarity": 0.025, - "size_similarity": 0.081, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.061, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 47210, - "right_pr_number": 47304, - "code_similarity": 0.042, - "size_similarity": 0.122, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.115, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 47301, - "right_pr_number": 47304, - "code_similarity": 0.143, - "size_similarity": 0.667, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.062, - "shared_filenames": [], - "shared_file_areas": [] - } - ] - }, - { - "cluster_id": "cluster-46658-2", - "summary": "Cluster of 2 related pull requests linked by soft_similarity.", - "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 46658, - "issue_numbers": [], "pr_numbers": [ - 46658, - 46836 + 43251, + 43254 ], "evidence_types": [ - "soft_similarity" + "closing_reference", + "shared_issue_target" ], "pr_comparisons": [ { - "left_pr_number": 46658, - "right_pr_number": 46836, - "code_similarity": 0.206, - "size_similarity": 0.71, - "file_overlap": 0.059, - "area_overlap": 0.0, - "patch_similarity": 0.308, + "left_pr_number": 43251, + "right_pr_number": 43254, + "code_similarity": 0.794, + "size_similarity": 0.64, + "file_overlap": 1.0, + "area_overlap": 0.667, + "patch_similarity": 0.882, "shared_filenames": [ - "src/transformers/conversion_mapping.py", - "src/transformers/models/auto/auto_mappings.py", - "src/transformers/models/auto/modeling_auto.py" + "src/transformers/loss/loss_utils.py" ], "shared_file_areas": [ { - "filename": "src/transformers/conversion_mapping.py", + "filename": "src/transformers/loss/loss_utils.py", "left_ranges": [ [ - 739, - 830 + 30, + 50 ] ], "right_ranges": [ [ - 127, - 133 - ], - [ - 1495, - 1545 - ] - ] - }, - { - "filename": "src/transformers/models/auto/auto_mappings.py", - "left_ranges": [ - [ - 362, - 368 - ], - [ - 591, - 599 - ], - [ - 798, - 804 - ], - [ - 884, - 891 - ], - [ - 1084, - 1090 - ], - [ - 1207, - 1213 - ] - ], - "right_ranges": [ - [ - 637, - 647 - ], - [ - 919, - 928 - ], - [ - 1130, - 1136 - ], - [ - 1260, - 1266 - ] - ] - }, - { - "filename": "src/transformers/models/auto/modeling_auto.py", - "left_ranges": [ - [ - 1085, - 1091 - ] - ], - "right_ranges": [ - [ - 509, - 515 - ], - [ - 1119, - 1125 - ] - ] - } - ] - } - ] - }, - { - "cluster_id": "cluster-43240-3", - "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", - "status": "open", - "confidence": 0.75, - "canonical_issue_number": 43240, - "canonical_pr_number": 43251, - "issue_numbers": [ - 43240 - ], - "pr_numbers": [ - 43251, - 43254 - ], - "evidence_types": [ - "closing_reference", - "shared_issue_target" - ], - "pr_comparisons": [ - { - "left_pr_number": 43251, - "right_pr_number": 43254, - "code_similarity": 0.794, - "size_similarity": 0.64, - "file_overlap": 1.0, - "area_overlap": 0.667, - "patch_similarity": 0.882, - "shared_filenames": [ - "src/transformers/loss/loss_utils.py" - ], - "shared_file_areas": [ - { - "filename": "src/transformers/loss/loss_utils.py", - "left_ranges": [ - [ - 30, - 50 - ] - ], - "right_ranges": [ - [ - 30, - 43 + 30, + 43 ] ] } @@ -2445,6 +2330,60 @@ } ] }, + { + "cluster_id": "cluster-46869-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46869.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 46869, + "canonical_pr_number": 46886, + "issue_numbers": [ + 46869 + ], + "pr_numbers": [ + 46886, + 46922 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 46886, + "right_pr_number": 46922, + "code_similarity": 0.711, + "size_similarity": 0.857, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.429, + "shared_filenames": [ + "src/transformers/models/yolos/modeling_yolos.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/yolos/modeling_yolos.py", + "left_ranges": [ + [ + 446, + 452 + ] + ], + "right_ranges": [ + [ + 434, + 440 + ], + [ + 446, + 452 + ] + ] + } + ] + } + ] + }, { "cluster_id": "cluster-45561-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #45561.", @@ -2536,6 +2475,58 @@ } ] }, + { + "cluster_id": "cluster-47096-3", + "summary": "Cluster of 3 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47096, + "issue_numbers": [], + "pr_numbers": [ + 47096, + 47306, + 47367 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47096, + "right_pr_number": 47306, + "code_similarity": 0.169, + "size_similarity": 0.78, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.089, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47096, + "right_pr_number": 47367, + "code_similarity": 0.025, + "size_similarity": 0.097, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.04, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47306, + "right_pr_number": 47367, + "code_similarity": 0.021, + "size_similarity": 0.076, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.041, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, { "cluster_id": "cluster-46305-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #46305.", @@ -2749,698 +2740,37 @@ ], "pr_comparisons": [ { - "left_pr_number": 43779, - "right_pr_number": 43816, - "code_similarity": 0.418, - "size_similarity": 0.538, - "file_overlap": 1.0, - "area_overlap": 0.02, - "patch_similarity": 0.02, - "shared_filenames": [ - "src/transformers/integrations/integration_utils.py" - ], - "shared_file_areas": [ - { - "filename": "src/transformers/integrations/integration_utils.py", - "left_ranges": [ - [ - 2227, - 2241 - ], - [ - 2303, - 2309 - ] - ], - "right_ranges": [ - [ - 2278, - 2291 - ], - [ - 2309, - 2322 - ] - ] - } - ] - } - ] - }, - { - "cluster_id": "cluster-47003-2", - "summary": "Cluster of 2 related pull requests linked by soft_similarity.", - "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 47243, - "issue_numbers": [], - "pr_numbers": [ - 47003, - 47243 - ], - "evidence_types": [ - "soft_similarity" - ], - "pr_comparisons": [ - { - "left_pr_number": 47003, - "right_pr_number": 47243, - "code_similarity": 0.206, - "size_similarity": 0.376, - "file_overlap": 0.093, - "area_overlap": 0.176, - "patch_similarity": 0.276, - "shared_filenames": [ - "src/transformers/exporters/configs.py", - "src/transformers/exporters/exporter_dynamo.py", - "src/transformers/exporters/exporter_executorch.py", - "src/transformers/exporters/exporter_onnx.py", - "src/transformers/exporters/utils.py", - "src/transformers/models/xcodec2/modeling_xcodec2.py", - "src/transformers/models/xcodec2/modular_xcodec2.py", - "tests/exporters/test_export.py" - ], - "shared_file_areas": [ - { - "filename": "src/transformers/exporters/configs.py", - "left_ranges": [ - [ - 28, - 34 - ], - [ - 100, - 107 - ], - [ - 142, - 147 - ], - [ - 168, - 200 - ] - ], - "right_ranges": [ - [ - 163, - 185 - ] - ] - }, - { - "filename": "src/transformers/exporters/exporter_dynamo.py", - "left_ranges": [ - [ - 43, - 49 - ], - [ - 352, - 359 - ], - [ - 383, - 390 - ], - [ - 452, - 473 - ], - [ - 490, - 510 - ], - [ - 551, - 562 - ], - [ - 597, - 609 - ] - ], - "right_ranges": [ - [ - 43, - 49 - ], - [ - 426, - 443 - ], - [ - 462, - 478 - ], - [ - 495, - 509 - ], - [ - 550, - 558 - ] - ] - }, - { - "filename": "src/transformers/exporters/exporter_executorch.py", - "left_ranges": [ - [ - 70, - 88 - ], - [ - 196, - 204 - ] - ], - "right_ranges": [ - [ - 19, - 29 - ], - [ - 40, - 46 - ], - [ - 52, - 57 - ], - [ - 62, - 75 - ], - [ - 80, - 99 - ], - [ - 142, - 157 - ], - [ - 184, - 207 - ], - [ - 211, - 263 - ], - [ - 273, - 285 - ], - [ - 303, - 334 - ], - [ - 351, - 363 - ], - [ - 382, - 529 - ], - [ - 548, - 554 - ], - [ - 559, - 601 - ], - [ - 608, - 623 - ], - [ - 654, - 692 - ], - [ - 706, - 737 - ], - [ - 748, - 753 - ], - [ - 808, - 813 - ], - [ - 841, - 846 - ], - [ - 864, - 988 - ], - [ - 1008, - 1013 - ], - [ - 1030, - 1373 - ], - [ - 1395, - 1418 - ], - [ - 1432, - 1440 - ], - [ - 1448, - 1462 - ], - [ - 1475, - 1516 - ], - [ - 1652, - 1704 - ] - ] - }, - { - "filename": "src/transformers/exporters/exporter_onnx.py", - "left_ranges": [ - [ - 74, - 100 - ], - [ - 146, - 152 - ], - [ - 210, - 221 - ], - [ - 270, - 310 - ], - [ - 365, - 390 - ], - [ - 497, - 504 - ], - [ - 749, - 758 - ], - [ - 829, - 837 - ], - [ - 865, - 875 - ], - [ - 898, - 914 - ] - ], - "right_ranges": [ - [ - 337, - 389 - ] - ] - }, - { - "filename": "src/transformers/exporters/utils.py", - "left_ranges": [ - [ - 232, - 238 - ], - [ - 432, - 441 - ], - [ - 444, - 467 - ], - [ - 472, - 498 - ], - [ - 514, - 536 - ], - [ - 662, - 706 - ], - [ - 747, - 759 - ], - [ - 808, - 815 - ] - ], - "right_ranges": [ - [ - 607, - 639 - ], - [ - 748, - 754 - ], - [ - 760, - 775 - ] - ] - }, - { - "filename": "src/transformers/models/xcodec2/modeling_xcodec2.py", - "left_ranges": [ - [ - 771, - 783 - ], - [ - 798, - 804 - ] - ], - "right_ranges": [ - [ - 776, - 782 - ] - ] - }, - { - "filename": "src/transformers/models/xcodec2/modular_xcodec2.py", - "left_ranges": [ - [ - 446, - 458 - ], - [ - 473, - 479 - ] - ], - "right_ranges": [ - [ - 451, - 457 - ] - ] - }, - { - "filename": "tests/exporters/test_export.py", - "left_ranges": [ - [ - 13, - 21 - ], - [ - 26, - 32 - ], - [ - 37, - 43 - ], - [ - 49, - 132 - ], - [ - 239, - 279 - ], - [ - 307, - 341 - ], - [ - 354, - 461 - ], - [ - 493, - 517 - ], - [ - 532, - 539 - ], - [ - 564, - 570 - ], - [ - 573, - 582 - ], - [ - 599, - 605 - ], - [ - 608, - 617 - ], - [ - 621, - 656 - ], - [ - 659, - 677 - ], - [ - 701, - 708 - ], - [ - 711, - 717 - ], - [ - 720, - 726 - ], - [ - 732, - 738 - ], - [ - 755, - 761 - ], - [ - 764, - 773 - ], - [ - 778, - 813 - ], - [ - 816, - 834 - ] - ], - "right_ranges": [ - [ - 13, - 39 - ], - [ - 112, - 134 - ], - [ - 155, - 248 - ], - [ - 292, - 322 - ], - [ - 327, - 358 - ], - [ - 373, - 474 - ], - [ - 512, - 528 - ], - [ - 531, - 546 - ], - [ - 552, - 561 - ], - [ - 590, - 596 - ], - [ - 599, - 605 - ], - [ - 625, - 631 - ], - [ - 656, - 664 - ], - [ - 668, - 688 - ], - [ - 701, - 757 - ], - [ - 767, - 797 - ], - [ - 804, - 838 - ] - ] - } - ] - } - ] - }, - { - "cluster_id": "cluster-47096-2", - "summary": "Cluster of 2 related pull requests linked by soft_similarity.", - "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 47096, - "issue_numbers": [], - "pr_numbers": [ - 47096, - 47383 - ], - "evidence_types": [ - "soft_similarity" - ], - "pr_comparisons": [ - { - "left_pr_number": 47096, - "right_pr_number": 47383, - "code_similarity": 0.103, - "size_similarity": 0.314, - "file_overlap": 0.125, - "area_overlap": 0.0, - "patch_similarity": 0.02, - "shared_filenames": [ - "tests/models/kimi_k25/test_modeling_kimi_k25.py" - ], - "shared_file_areas": [ - { - "filename": "tests/models/kimi_k25/test_modeling_kimi_k25.py", - "left_ranges": [ - [ - 127, - 132 - ] - ], - "right_ranges": [ - [ - 18, - 24 - ], - [ - 27, - 36 - ], - [ - 216, - 401 - ] - ] - } - ] - } - ] - }, - { - "cluster_id": "cluster-46912-2", - "summary": "Cluster of 2 related pull requests linked by soft_similarity.", - "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 47175, - "issue_numbers": [], - "pr_numbers": [ - 46912, - 47175 - ], - "evidence_types": [ - "soft_similarity" - ], - "pr_comparisons": [ - { - "left_pr_number": 46912, - "right_pr_number": 47175, - "code_similarity": 0.011, - "size_similarity": 0.005, - "file_overlap": 0.029, - "area_overlap": 0.0, - "patch_similarity": 0.01, + "left_pr_number": 43779, + "right_pr_number": 43816, + "code_similarity": 0.418, + "size_similarity": 0.538, + "file_overlap": 1.0, + "area_overlap": 0.02, + "patch_similarity": 0.02, "shared_filenames": [ - "tests/models/olmo3/test_modeling_olmo3.py" + "src/transformers/integrations/integration_utils.py" ], "shared_file_areas": [ { - "filename": "tests/models/olmo3/test_modeling_olmo3.py", + "filename": "src/transformers/integrations/integration_utils.py", "left_ranges": [ [ - 99, - 104 + 2227, + 2241 + ], + [ + 2303, + 2309 ] ], "right_ranges": [ [ - 213, - 227 + 2278, + 2291 ], [ - 312, - 318 + 2309, + 2322 ] ] } @@ -3449,29 +2779,29 @@ ] }, { - "cluster_id": "cluster-47163-2", + "cluster_id": "cluster-47175-2", "summary": "Cluster of 2 related pull requests linked by soft_similarity.", "status": "open", "confidence": 0.5, "canonical_issue_number": null, - "canonical_pr_number": 47166, + "canonical_pr_number": 47298, "issue_numbers": [], "pr_numbers": [ - 47163, - 47166 + 47175, + 47298 ], "evidence_types": [ "soft_similarity" ], "pr_comparisons": [ { - "left_pr_number": 47163, - "right_pr_number": 47166, - "code_similarity": 0.104, - "size_similarity": 0.483, + "left_pr_number": 47175, + "right_pr_number": 47298, + "code_similarity": 0.019, + "size_similarity": 0.043, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.047, + "patch_similarity": 0.067, "shared_filenames": [], "shared_file_areas": [] } @@ -3608,31 +2938,305 @@ ] }, { - "cluster_id": "cluster-47173-2", + "cluster_id": "cluster-47443-2", "summary": "Cluster of 2 related pull requests linked by soft_similarity.", "status": "open", "confidence": 0.5, "canonical_issue_number": null, - "canonical_pr_number": 47292, + "canonical_pr_number": 47444, "issue_numbers": [], "pr_numbers": [ - 47173, - 47292 + 47443, + 47444 ], "evidence_types": [ "soft_similarity" ], "pr_comparisons": [ { - "left_pr_number": 47173, - "right_pr_number": 47292, - "code_similarity": 0.097, - "size_similarity": 0.463, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.027, - "shared_filenames": [], - "shared_file_areas": [] + "left_pr_number": 47443, + "right_pr_number": 47444, + "code_similarity": 0.754, + "size_similarity": 0.632, + "file_overlap": 0.818, + "area_overlap": 1.0, + "patch_similarity": 0.217, + "shared_filenames": [ + "src/transformers/models/deepseek_v3/modeling_deepseek_v3.py", + "src/transformers/models/deepseek_v32/modeling_deepseek_v32.py", + "src/transformers/models/glm4_moe_lite/modeling_glm4_moe_lite.py", + "src/transformers/models/glm_moe_dsa/modeling_glm_moe_dsa.py", + "src/transformers/models/longcat_flash/modeling_longcat_flash.py", + "src/transformers/models/minicpm3/modeling_minicpm3.py", + "src/transformers/models/mistral4/modeling_mistral4.py", + "src/transformers/models/nemotron_h/modeling_nemotron_h.py", + "src/transformers/models/youtu/modeling_youtu.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/deepseek_v3/modeling_deepseek_v3.py", + "left_ranges": [ + [ + 4, + 9 + ], + [ + 310, + 315 + ] + ], + "right_ranges": [ + [ + 4, + 9 + ], + [ + 310, + 315 + ] + ] + }, + { + "filename": "src/transformers/models/deepseek_v32/modeling_deepseek_v32.py", + "left_ranges": [ + [ + 299, + 304 + ], + [ + 338, + 349 + ], + [ + 392, + 403 + ] + ], + "right_ranges": [ + [ + 299, + 304 + ], + [ + 338, + 349 + ], + [ + 392, + 403 + ] + ] + }, + { + "filename": "src/transformers/models/glm4_moe_lite/modeling_glm4_moe_lite.py", + "left_ranges": [ + [ + 182, + 187 + ], + [ + 221, + 232 + ], + [ + 272, + 283 + ] + ], + "right_ranges": [ + [ + 182, + 187 + ], + [ + 221, + 232 + ], + [ + 272, + 283 + ] + ] + }, + { + "filename": "src/transformers/models/glm_moe_dsa/modeling_glm_moe_dsa.py", + "left_ranges": [ + [ + 306, + 311 + ], + [ + 358, + 369 + ] + ], + "right_ranges": [ + [ + 306, + 311 + ], + [ + 358, + 369 + ] + ] + }, + { + "filename": "src/transformers/models/longcat_flash/modeling_longcat_flash.py", + "left_ranges": [ + [ + 282, + 287 + ], + [ + 321, + 332 + ], + [ + 372, + 383 + ] + ], + "right_ranges": [ + [ + 282, + 287 + ], + [ + 321, + 332 + ], + [ + 372, + 383 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modeling_minicpm3.py", + "left_ranges": [ + [ + 196, + 201 + ], + [ + 283, + 288 + ] + ], + "right_ranges": [ + [ + 196, + 201 + ], + [ + 283, + 288 + ] + ] + }, + { + "filename": "src/transformers/models/mistral4/modeling_mistral4.py", + "left_ranges": [ + [ + 17, + 22 + ], + [ + 319, + 324 + ] + ], + "right_ranges": [ + [ + 17, + 22 + ], + [ + 319, + 324 + ] + ] + }, + { + "filename": "src/transformers/models/nemotron_h/modeling_nemotron_h.py", + "left_ranges": [ + [ + 109, + 114 + ], + [ + 291, + 300 + ], + [ + 361, + 370 + ], + [ + 396, + 409 + ], + [ + 435, + 444 + ] + ], + "right_ranges": [ + [ + 109, + 114 + ], + [ + 291, + 300 + ], + [ + 361, + 370 + ], + [ + 396, + 409 + ], + [ + 435, + 444 + ] + ] + }, + { + "filename": "src/transformers/models/youtu/modeling_youtu.py", + "left_ranges": [ + [ + 222, + 227 + ], + [ + 261, + 272 + ], + [ + 312, + 323 + ] + ], + "right_ranges": [ + [ + 222, + 227 + ], + [ + 261, + 272 + ], + [ + 312, + 323 + ] + ] + } + ] } ] }, @@ -3914,12 +3518,13 @@ ], "duplicate_issues": [ { - "cluster_id": "cluster-46894-6", + "cluster_id": "cluster-46894-7", "canonical_issue_number": 46894, "duplicate_issue_numbers": [ - 46897 + 46897, + 46901 ], - "reason": "Issues in cluster-46894-6 are treated as duplicates because they share closing_reference, shared_issue_target, soft_similarity evidence." + "reason": "Issues in cluster-46894-7 are treated as duplicates because they share closing_reference, shared_issue_target, soft_similarity evidence." }, { "cluster_id": "cluster-47138-4", @@ -3928,6 +3533,14 @@ 47139 ], "reason": "Issues in cluster-47138-4 are treated as duplicates because they share closing_reference, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47332-4", + "canonical_issue_number": 47333, + "duplicate_issue_numbers": [ + 47332 + ], + "reason": "Issues in cluster-47332-4 are treated as duplicates because they share closing_reference, soft_similarity evidence." } ], "duplicate_prs": [ @@ -3950,16 +3563,13 @@ "reason": "PRs in cluster-43240-3 are treated as duplicates because they converge on issue #43240 with closing_reference, shared_issue_target evidence." }, { - "cluster_id": "cluster-43665-6", - "canonical_pr_number": 46886, + "cluster_id": "cluster-43451-3", + "canonical_pr_number": 47322, "duplicate_pr_numbers": [ - 43665, - 44815, - 46380, - 46922 + 43451 ], - "target_issue_number": 46869, - "reason": "PRs in cluster-43665-6 are treated as duplicates because they converge on issue #46869 with closing_reference, shared_issue_target, soft_similarity evidence." + "target_issue_number": 47321, + "reason": "PRs in cluster-43451-3 are treated as duplicates because they converge on issue #47321 with closing_reference, soft_similarity evidence." }, { "cluster_id": "cluster-43698-3", @@ -4042,16 +3652,28 @@ "reason": "PRs in cluster-46305-3 are treated as duplicates because they converge on issue #46305 with closing_reference, shared_issue_target, soft_similarity evidence." }, { - "cluster_id": "cluster-46658-2", - "canonical_pr_number": 46658, + "cluster_id": "cluster-46509-5", + "canonical_pr_number": 47143, "duplicate_pr_numbers": [ - 46836 + 46509, + 47163, + 47301, + 47387 ], "target_issue_number": null, - "reason": "PRs in cluster-46658-2 are treated as duplicates because they share soft_similarity evidence." + "reason": "PRs in cluster-46509-5 are treated as duplicates because they share soft_similarity evidence." }, { - "cluster_id": "cluster-46894-6", + "cluster_id": "cluster-46869-3", + "canonical_pr_number": 46886, + "duplicate_pr_numbers": [ + 46922 + ], + "target_issue_number": 46869, + "reason": "PRs in cluster-46869-3 are treated as duplicates because they converge on issue #46869 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-46894-7", "canonical_pr_number": 46903, "duplicate_pr_numbers": [ 46895, @@ -4059,45 +3681,17 @@ 47090 ], "target_issue_number": 46894, - "reason": "PRs in cluster-46894-6 are treated as duplicates because they converge on issue #46894 with closing_reference, shared_issue_target, soft_similarity evidence." - }, - { - "cluster_id": "cluster-46912-2", - "canonical_pr_number": 47175, - "duplicate_pr_numbers": [ - 46912 - ], - "target_issue_number": null, - "reason": "PRs in cluster-46912-2 are treated as duplicates because they share soft_similarity evidence." - }, - { - "cluster_id": "cluster-47003-2", - "canonical_pr_number": 47243, - "duplicate_pr_numbers": [ - 47003 - ], - "target_issue_number": null, - "reason": "PRs in cluster-47003-2 are treated as duplicates because they share soft_similarity evidence." - }, - { - "cluster_id": "cluster-47077-4", - "canonical_pr_number": 47077, - "duplicate_pr_numbers": [ - 47210, - 47301, - 47304 - ], - "target_issue_number": null, - "reason": "PRs in cluster-47077-4 are treated as duplicates because they share soft_similarity evidence." + "reason": "PRs in cluster-46894-7 are treated as duplicates because they converge on issue #46894 with closing_reference, shared_issue_target, soft_similarity evidence." }, { - "cluster_id": "cluster-47096-2", + "cluster_id": "cluster-47096-3", "canonical_pr_number": 47096, "duplicate_pr_numbers": [ - 47383 + 47306, + 47367 ], "target_issue_number": null, - "reason": "PRs in cluster-47096-2 are treated as duplicates because they share soft_similarity evidence." + "reason": "PRs in cluster-47096-3 are treated as duplicates because they share soft_similarity evidence." }, { "cluster_id": "cluster-47123-2", @@ -4118,22 +3712,13 @@ "reason": "PRs in cluster-47138-4 are treated as duplicates because they converge on issue #47138 with closing_reference, soft_similarity evidence." }, { - "cluster_id": "cluster-47163-2", - "canonical_pr_number": 47166, + "cluster_id": "cluster-47175-2", + "canonical_pr_number": 47298, "duplicate_pr_numbers": [ - 47163 - ], - "target_issue_number": null, - "reason": "PRs in cluster-47163-2 are treated as duplicates because they share soft_similarity evidence." - }, - { - "cluster_id": "cluster-47173-2", - "canonical_pr_number": 47292, - "duplicate_pr_numbers": [ - 47173 + 47175 ], "target_issue_number": null, - "reason": "PRs in cluster-47173-2 are treated as duplicates because they share soft_similarity evidence." + "reason": "PRs in cluster-47175-2 are treated as duplicates because they share soft_similarity evidence." }, { "cluster_id": "cluster-47282-2", @@ -4144,6 +3729,15 @@ "target_issue_number": null, "reason": "PRs in cluster-47282-2 are treated as duplicates because they share soft_similarity evidence." }, + { + "cluster_id": "cluster-47332-4", + "canonical_pr_number": 47336, + "duplicate_pr_numbers": [ + 47383 + ], + "target_issue_number": 47333, + "reason": "PRs in cluster-47332-4 are treated as duplicates because they converge on issue #47333 with closing_reference, soft_similarity evidence." + }, { "cluster_id": "cluster-47429-3", "canonical_pr_number": 47434, @@ -4152,13 +3746,22 @@ ], "target_issue_number": 47429, "reason": "PRs in cluster-47429-3 are treated as duplicates because they converge on issue #47429 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47443-2", + "canonical_pr_number": 47444, + "duplicate_pr_numbers": [ + 47443 + ], + "target_issue_number": null, + "reason": "PRs in cluster-47443-2 are treated as duplicates because they share soft_similarity evidence." } ], "best_issue": { - "cluster_id": "cluster-43665-6", - "issue_number": 46869, - "reason": "Issue #46869 is the strongest global issue candidate because it is open, belongs to a cluster with 6 artifacts, and carries 2 discussion comments plus 6 inbound references.", - "score": 170.0 + "cluster_id": "cluster-46894-7", + "issue_number": 46894, + "reason": "Issue #46894 is the strongest global issue candidate because it is open, belongs to a cluster with 7 artifacts, and carries 1 discussion comments plus 10 inbound references.", + "score": 182.0 }, "best_pr": { "cluster_id": "cluster-43979-11", diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index 4eb8e01d5ccb8b181a33d6695935a2029a84c2fd..fdabdef7912f35457f37d6186120b6c4c5941688 100644 --- a/analysis/current/analysis-report-hybrid.llm-reviews.json +++ b/analysis/current/analysis-report-hybrid.llm-reviews.json @@ -1,8 +1,8 @@ { "schema_version": "1.0", "repo": "huggingface/transformers", - "snapshot_id": "20260721T000056Z", - "generated_at": "2026-07-21T00:24:00Z", + "snapshot_id": "20260721T060018Z", + "generated_at": "2026-07-21T06:15:09Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -125,20 +125,20 @@ "issue:43039|issue:44918", "issue:41762|issue:43638" ], - "prepared_review_unit_hash": "c568a309364b7148b863644a98b5dab31bcde93c7b2cb25c2c220070fa02d343", + "prepared_review_unit_hash": "1b6f1566d5de3821abc767cf96c50cf01873b44348a87cfc903e0678e17740c8", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 56, - "serialized_chars": 44518, + "serialized_chars": 44520, "estimated_input_tokens": 11130, "estimated_eval_tokens": 22516 }, @@ -147,357 +147,297 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most items are unrelated single issues spanning different models, runtime setups, and docs regressions. The only clear duplicate cluster is the Deepspeed ZeRO-3 + BertModel index-out-of-bounds crash; the torch 2.4.1 compatibility reports also look like the same underlying incompatibility.", - "confidence": 0.79, - "canonical_issue_reason": "issue:43596 is the clearest representative of the Bert + ZeRO-3 index error, with a broad title and a concrete failure mode that matches the duplicate report.", + "summary": "Mostly heterogeneous training/model bugs, with two credible duplicate pockets: ZeRO-3 BERT init (43596/43638) and Qwen3_5Moe rotary embedding materialization (45859/45902). The torch 2.4.1 / transformers pair (45800/46515) is too weakly evidenced to call an exact duplicate.", + "confidence": 0.71, + "canonical_issue_reason": "Issue 43638 is the strongest exact-duplicate anchor: 43596 describes the same ZeRO-3 BERT empty-tensor / index-out-of-bounds initialization failure.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 is the best canonical issue because it describes the shared crash pattern succinctly and is broad enough to subsume the near-duplicate BertModel report.", + "best_issue_reason": "43638 is the best global issue representative because it anchors the clearest confirmed duplicate pocket, while most of the surrounding cluster is unrelated.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different models and failures: a missing config attribute in Qwen3_5Config vs streaming chat/completions failure in Qwen3.5-0.8B." + "reason": "Both involve Qwen3.5, but one is a missing config attribute and the other is a streaming chat/completions failure; different bugs." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Unrelated hardware/runtime failures: AMD GPU support for GPT-OSS-20B vs an MPS histogram kernel not implemented for Int." + "reason": "Different models and failure modes: GPT-OSS AMD GPU support vs an MPS histogram kernel error." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different bug class and code path: image preprocessor/model dtype mismatch vs GPT-OSS-20B failing on AMD GPUs." + "reason": "Different model stacks and symptoms; a dtype mismatch in Mistral-3 is not the same bug as GPT-OSS failing on AMD GPUs." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Both mention runtime issues, but one is an autocast dtype mismatch and the other is an AMD GPU compatibility problem." + "reason": "Phi-tiny-MoE autocast dtype mismatch and GPT-OSS AMD GPU failure are unrelated." }, { "left": "issue:45237", "right": "issue:46468", "accept": false, - "reason": "Different models and symptoms: GPT-OSS GPU support vs a broken official Nemotron example." + "reason": "Both are runtime failures, but they concern different models and code paths, not the same underlying bug." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Not the same problem: GPT-OSS AMD GPU failure vs GraniteMoEHybrid invoking an invalid method." + "reason": "GPT-OSS AMD GPU failure and GraniteMoEHybrid invalid method call are unrelated." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "Different subsystems: DeepSpeed ZeRO-3 collective mismatch in Jamba vs ModernBERT ONNX export error." + "reason": "DeepSpeed ZeRO3 collective mismatch is unrelated to a ModernBERT ONNX export error." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Both are model-specific failures, but ONNX export for ModernBERT and Qwen3Moe FP8 failure are unrelated." + "reason": "ONNX export failure and FineGrainedFP8Config failure are different model/runtime problems." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "Different models and runtime contexts: ModernBERT export error vs Qwen3 auto device mapping cuda assert." + "reason": "ModernBERT ONNX export and Qwen3 auto device map cuda assert are unrelated failures." }, { "left": "issue:45901", "right": "issue:46015", "accept": false, - "reason": "Docs/example maintenance is the only commonality; table-question-answering crash and RAG examples are not the same bug." + "reason": "Table QA crash and outdated RAG docs are different issue types and different fixes." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "Different failures: ONNX export vs import-time GenerationMixin rename/removal issue." + "reason": "ONNX export error is unrelated to a missing GenerationMixin import." }, { "left": "issue:44918", "right": "issue:46165", "accept": false, - "reason": "Different training flows and symptoms: Qwen3.5 input embedding unpacking vs resuming local-model training with trust_remote_code." + "reason": "Qwen3.5 embedding unpacking under TRL SFT trainer is not the same as resume-training with trust_remote_code." }, { "left": "issue:46165", "right": "issue:46869", "accept": false, - "reason": "Unrelated problems: remote-code resume flow vs YOLO multi-GPU device_map auto failure." + "reason": "Resume-training trust_remote_code and YOLO multi-GPU device_map issues are unrelated." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are stale v5 docs issues, but they target different removed items and are not the same concrete documentation defect." + "reason": "Both are v5 docs regressions, but they reference different removed docs entries and different cleanup work." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Different model families and errors: mask/tensor shape mismatch vs Gemma4 PLE device mismatch." + "reason": "Mask shape mismatch is unrelated to Gemma4 PLE device mismatch." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Different failure modes: ZeRO-3/Bert index-out-of-bounds vs a mask shape mismatch." + "reason": "ZeRO-3 BERT init failure and a mask/indexing shape mismatch are different bugs." }, { "left": "issue:43638", "right": "issue:46165", "accept": false, - "reason": "Bert ZeRO-3 initialization crash and remote-code training resume issue are unrelated." + "reason": "ZeRO-3 BERT init failure is unrelated to the resume-training trust_remote_code issue." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Different root causes: BertModel ZeRO-3 index error vs a mask/tensor indexing shape mismatch." + "reason": "Both are tensor/indexing errors, but the affected models and code paths differ." }, { "left": "issue:44805", "right": "issue:46294", "accept": false, - "reason": "Different modalities and bugs: generic mask shape error vs Qwen3.5 video token/feature mismatch." + "reason": "Mask/index shape mismatch is not the same as Qwen3.5 video token / feature mismatch." }, { "left": "issue:46165", "right": "issue:46468", "accept": false, - "reason": "No shared concrete code path; these are separate training/resume and example-breakage issues." - }, - { - "left": "issue:44805", - "right": "issue:45161", - "accept": false, - "reason": "Different model/runtime bugs: mask shape mismatch vs GPT-OSS MoE tensor-parallel issue." + "reason": "Resume-training with remote code and a broken Nemotron example are unrelated." }, { "left": "issue:44560", "right": "issue:46165", "accept": false, - "reason": "Different features and failures: Qwen3-vl video input StopIteration vs model-training resume with remote code." + "reason": "Qwen3-vl video StopIteration and resume-training trust_remote_code are different bugs." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "Unrelated: BarkProcessor voice preset handling vs ModernBERT export failure." + "reason": "BarkProcessor voice_preset and ModernBERT ONNX export are unrelated." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "Both involve DeepSpeed, but the underlying problems differ: collective mismatch vs Bert zero-size index error." - }, - { - "left": "issue:43596", - "right": "issue:43638", - "accept": true, - "reason": "Same underlying ZeRO-3/BertModel bug: both report an index-out-of-bounds on size-0 tensors during DeepSpeed ZeRO-3 initialization, differing only in model-pretraining details." - }, - { - "left": "issue:43039", - "right": "issue:45161", - "accept": false, - "reason": "Liger Kernel cross-entropy invocation and GPT-OSS tensor-parallel failure are distinct issues." - }, - { - "left": "issue:43975", - "right": "issue:44457", - "accept": false, - "reason": "Different problems: DeepSeek coder detokenization regression vs LoRA merge/save/load output mismatch." - }, - { - "left": "issue:34928", - "right": "issue:35308", - "accept": false, - "reason": "Both affect distributed training, but activation-checkpointing/FSDP tensor recomputation and IterableDataset length mismatch are different bugs." - }, - { - "left": "issue:44292", - "right": "issue:44457", - "accept": false, - "reason": "Different domains: Qwen-3-8B-NVFP4 runtime error vs LoRA merge/reload output inconsistency." - }, - { - "left": "issue:44841", - "right": "issue:46661", - "accept": false, - "reason": "Different layers of the stack: Voxtral processor failure vs remote-code AutoModel config_class attribute error." - }, - { - "left": "issue:24643", - "right": "issue:30277", - "accept": false, - "reason": "Different DeepSpeed failures: 2-D weight training error vs rank collective mismatch in Jamba-ZeRO-3." + "reason": "Both involve DeepSpeed, but the concrete failures are different and not mergeable into one fix." }, { "left": "issue:43716", "right": "issue:46869", "accept": false, - "reason": "Different device/dtype class problems: Mistral image preprocessor mismatch vs YOLO multi-GPU device_map auto failure." + "reason": "Mistral-3 dtype mismatch and YOLO multi-GPU device map failure are unrelated." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Both are tensor-shape-related, but one is activation checkpointing/FSDP recomputation and the other is a mask indexing mismatch." + "reason": "FSDP activation-checkpointing mismatch and mask indexing shape mismatch are different distributed-training bugs." }, { "left": "issue:45685", "right": "issue:46869", "accept": false, - "reason": "MPS histogram kernel limitation and YOLO multi-GPU auto device mapping are unrelated." + "reason": "MPS histogram kernel error and YOLO device_map auto failure are unrelated." }, { "left": "issue:34928", "right": "issue:36331", "accept": false, - "reason": "Different training stack changes: FSDP/checkpointing shape mismatch vs Trainer API signature change for compute_loss." + "reason": "FSDP activation-checkpointing mismatch is unrelated to a Trainer API signature change." }, { "left": "issue:45859", "right": "issue:45902", - "accept": false, - "reason": "Same model component family, but different concrete bugs: CPU offload incompatibility vs uninitialized-memory after materialization." + "accept": true, + "reason": "Same Qwen3_5MoeTextRotaryEmbedding path; both indicate broken device/materialization handling of rotary embedding state under offload/meta-to-empty flows." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "Different failures and models: ZeRO-3 Bert index error vs GPT-OSS MoE tensor-parallel issue." + "reason": "ZeRO-3 BERT init failure and GPT-OSS tensor-parallel failure are different bugs." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "No shared code path: Bert ZeRO-3 initialization crash vs GPT-OSS MoE TP failure." + "reason": "Same as above: different model and failure mode." }, { "left": "issue:43716", "right": "issue:45507", "accept": false, - "reason": "Different model families and symptoms: Mistral dtype mismatch vs GraniteMoEHybrid invalid method call." + "reason": "Mistral-3 dtype mismatch and GraniteMoEHybrid invalid method call are unrelated." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Different models and environments: Qwen-3-8B-NVFP4 vs GPT-OSS-20B on AMD GPUs." + "reason": "Qwen NVFP4 runtime error and GPT-OSS AMD GPU failure are unrelated." }, { "left": "issue:45874", "right": "issue:46294", "accept": false, - "reason": "Both are Gemma/Qwen multimodal issues, but expensive reverse-embedding lookup and video token mismatch are different problems." + "reason": "Gemma4 reverse embedding lookup and Qwen3.5 video token mismatch are different bugs." }, { "left": "issue:39290", "right": "issue:45081", "accept": false, - "reason": "Unrelated config regression and tokenizer patch crash." + "reason": "Gemma3TextConfig missing attribute and Mistral tokenizer regex patch crash are unrelated." }, { "left": "issue:45800", "right": "issue:46515", - "accept": true, - "reason": "These appear to be the same torch 2.4.1 compatibility regression; one report just adds the Transformers 5.8.0 context." + "accept": false, + "reason": "The titles are similar, but 46515 only says torch 2.4.1 incompatibility and does not establish the same transformers 5.8.0-specific bug or fix path." }, { "left": "issue:34928", "right": "issue:41720", "accept": false, - "reason": "Different distributed-training failures: recomputed tensor-size mismatch vs cuda assert under auto device mapping." + "reason": "FSDP activation checkpointing mismatch and Qwen3 cuda assert are different failures." }, { "left": "issue:45823", "right": "issue:45874", "accept": false, - "reason": "Both involve Gemma4, but device mismatch during forward and reverse-embedding lookup on inputs_embeds are different concrete bugs." + "reason": "Gemma4 PLE device mismatch and Gemma4 inputs_embeds reverse lookup are separate issues." }, { "left": "issue:43425", "right": "issue:44292", "accept": false, - "reason": "General torch incompatibility vs a Qwen-3-8B-NVFP4 runtime error are not the same issue." + "reason": "Torch 2.10 incompatibility and Qwen-3-8B-NVFP4 error are different version/model issues." }, { "left": "issue:43828", "right": "issue:45161", "accept": false, - "reason": "Autocast dtype mismatch and GPT-OSS MoE tensor-parallel failure are different code paths." + "reason": "Phi-tiny-MoE autocast dtype mismatch is unrelated to GPT-OSS tensor-parallel failure." }, { "left": "issue:43541", "right": "issue:43716", "accept": false, - "reason": "Different failure modes and models: torch dynamo tracing error in Mixtral vs Mistral image preprocessor dtype mismatch." + "reason": "Mixtral grouped_mm torch dynamo tracing failure and Mistral-3 dtype mismatch are different bugs." }, { "left": "issue:43828", "right": "issue:44292", "accept": false, - "reason": "Different models and failure surfaces: autocast dtype mismatch vs Qwen-3-8B-NVFP4 runtime error." + "reason": "Different models and failures: autocast dtype mismatch vs NVFP4 runtime error." }, { "left": "issue:43716", "right": "issue:45685", "accept": false, - "reason": "Different hardware/runtime issues: dtype mismatch in Mistral multimodal processing vs an MPS histogram kernel limitation." + "reason": "Mistral-3 image preprocessor dtype mismatch and MPS histogram kernel error are unrelated." }, { "left": "issue:45507", "right": "issue:45685", "accept": false, - "reason": "Invalid method call in GraniteMoEHybrid is unrelated to an MPS Int kernel limitation." + "reason": "GraniteMoEHybrid invalid method and MPS histogram kernel error are unrelated." }, { "left": "issue:44292", "right": "issue:44918", "accept": false, - "reason": "Different model families and failure points: Qwen-3-8B-NVFP4 runtime error vs Qwen3.5 input-embedding unpacking with TRL SFT trainer." - }, - { - "left": "issue:43716", - "right": "issue:44292", - "accept": false, - "reason": "Mistral dtype mismatch and Qwen-3-8B-NVFP4 runtime failure are unrelated." - }, - { - "left": "issue:44918", - "right": "issue:45161", - "accept": false, - "reason": "Different stack layers: TRL SFT input-embedding unpacking vs GPT-OSS MoE tensor-parallel issue." + "reason": "Qwen NVFP4 runtime error and TRL SFT embedding unpacking are different issues." }, { "left": "issue:43039", "right": "issue:44918", "accept": false, - "reason": "Liger Kernel cross-entropy call and Qwen3.5 embedding unpacking are distinct bugs." + "reason": "Liger Kernel cross_entropy usage and Qwen3.5 embedding unpacking are unrelated." }, { "left": "issue:41762", "right": "issue:43638", "accept": false, - "reason": "Gemma3 ZeRO-3 index error and Bert ZeRO-3 index error are similar in shape but involve different models and are not clearly the same bug." + "reason": "Gemma3 ZeRO-3 load failure and BERT ZeRO-3 init failure are only broadly similar; the concrete bugs differ." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The two accepted duplicates are well-supported by the titles/body patterns (43596/43638 and 45800/46515), and the remaining soft-edge rejections do not overstate similarity. The canonical issue choice and confidence also look reasonable." + "accept": false, + "feedback": "The overall clustering read is conservative and mostly grounded, but one soft-edge accept is too strong for the packet. In particular, issue:45800 vs issue:46515 is described as an exact duplicate even though the packet only shows similar incompatibility titles; 46515 does not explicitly mention transformers 5.8.0, so 'exact same' overstates the evidence. The Qwen3.5 rotary-embedding pair is more defensible as a likely duplicate, and most other rejections look appropriately cautious, but the overconfident 45800/46515 verdict means I would not accept this review as-is." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -575,8 +515,8 @@ "issue:43296|issue:46304", "issue:39290|issue:41720", "issue:43866|issue:46402", - "issue:45237|issue:45901", "issue:44589|issue:46304", + "issue:45237|issue:45901", "issue:41762|issue:43596", "issue:43039|issue:44560", "issue:43901|issue:44512", @@ -589,247 +529,247 @@ "issue:35308|issue:42617", "issue:45812|issue:45920", "issue:43872|issue:46304", - "issue:46901|issue:47394", - "issue:44292|issue:46015", "issue:39290|issue:43531", + "issue:44292|issue:46015", + "issue:46901|issue:47394", "issue:43866|issue:44863", "issue:41093|issue:43638", "issue:43792|issue:45901", "issue:43425|issue:45070", "issue:43854|issue:43866" ], - "prepared_review_unit_hash": "a0e252fa9654b0b95279fa138448b14f453280684486e92188c45626be349be7", + "prepared_review_unit_hash": "acdfc73bf66d72ebcab8bdf0dfae14b3bb4f794f2178ea915420c5f575e91b64", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 33, - "serialized_chars": 36959, - "estimated_input_tokens": 9240, - "estimated_eval_tokens": 18736 + "serialized_chars": 36962, + "estimated_input_tokens": 9241, + "estimated_eval_tokens": 18738 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most pairs are not duplicates; only two soft edges look like the same underlying bug family: ZeRO-3 model loading index errors, and the v5 AutoTokenizer wrong-token-ID regression across models. No PRs were provided.", - "confidence": 0.72, - "canonical_issue_reason": "issue:45920 is the best umbrella issue: it describes the broad v5 AutoTokenizer wrong-token-ID regression across multiple model families, making it the most representative target over the narrower Granite-only report in issue:45812.", + "summary": "The items split into mostly unrelated issues, with two clear duplicate subclusters: a v5 AutoTokenizer wrong-token-ID regression and a DeepSpeed ZeRO-3 zero-size loading/index error. I only accepted soft edges where the titles describe the same concrete bug; most others are broad subsystem similarities, not true duplicates.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45920 is the broadest and most representative report of the tokenizer regression, covering multiple models rather than a single Granite-only case.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 has the widest blast radius and clearest regression framing, so it is the most useful issue to keep as the cluster anchor.", + "best_issue_reason": "issue:45920 best captures the shared underlying v5 AutoTokenizer bug and is the strongest canonical issue for the accepted duplicate family.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43827", "right": "issue:44512", "accept": false, - "reason": "Both are documentation complaints, but they cover different docs and different removed APIs; not the same underlying change." + "reason": "Both are docs bugs, but they concern different removed commands and different pages." }, { "left": "issue:43828", "right": "issue:45685", "accept": false, - "reason": "Different failures on different backends/models: autocast dtype mismatch vs MPS histogram int support." + "reason": "Unrelated runtime errors: autocast dtype mismatch vs MPS histogram_int limitation." }, { "left": "issue:43039", "right": "issue:43828", "accept": false, - "reason": "Unrelated model/runtime errors; one is about Liger cross-entropy dispatch, the other about autocast dtype mismatch." + "reason": "Different bugs in different code paths: Liger cross-entropy vs autocast dtype mismatch." }, { "left": "issue:30277", "right": "issue:36331", "accept": false, - "reason": "Different code paths: distributed collective mismatch vs a trainer signature change." + "reason": "Different failures: collective mismatch under ZeRO-3 vs trainer API kwarg mismatch." }, { "left": "issue:34928", "right": "issue:36010", "accept": false, - "reason": "Activation checkpointing/FSDP tensor-size mismatch is unrelated to the GenerationMixin import error." + "reason": "Activation-checkpoint/FSDP size mismatch and GenerationMixin import error are unrelated." }, { "left": "issue:45237", "right": "issue:46015", "accept": false, - "reason": "AMD GPU runtime failure and outdated RAG docs are unrelated." + "reason": "GPU backend support issue vs documentation maintenance; not the same bug." }, { "left": "issue:43039", "right": "issue:44805", "accept": false, - "reason": "Different bugs: Liger kernel cross-entropy use vs mask/tensor shape mismatch." + "reason": "Different concrete failures: kernel call behavior vs mask/tensor shape mismatch." }, { "left": "issue:45800", "right": "issue:46304", "accept": false, - "reason": "Both are torch/transformers compatibility issues, but they are different concrete incompatibilities and code paths." + "reason": "Both mention version compatibility, but the affected environment and failure mode differ." }, { "left": "issue:43296", "right": "issue:46304", "accept": false, - "reason": "Different problem domains: vLLM model loading vs distributed build/runtime compatibility." + "reason": "vLLM model loading failure is unrelated to the distributed-support regression." }, { "left": "issue:39290", "right": "issue:41720", "accept": false, - "reason": "Both involve model execution failures, but the affected models, backends, and symptoms differ." + "reason": "Different models and different symptoms: missing config attr vs CUDA assert." }, { "left": "issue:43866", "right": "issue:46402", "accept": false, - "reason": "Checkpoint corruption / load failure vs missing text-encoder weight loading are not the same bug." + "reason": "Checkpoint corruption and weights-not-loaded are distinct model-loading problems." }, { - "left": "issue:45237", - "right": "issue:45901", + "left": "issue:44589", + "right": "issue:46304", "accept": false, - "reason": "Different application areas and failure modes; no shared concrete code path." + "reason": "Storage-class lookup failure is unrelated to USE_DISTRIBUTED=0 support." }, { - "left": "issue:44589", - "right": "issue:46304", + "left": "issue:45237", + "right": "issue:45901", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to USE_DISTRIBUTED compatibility." + "reason": "AMD GPU incompatibility and table QA crash are unrelated." }, { "left": "issue:41762", "right": "issue:43596", "accept": true, - "reason": "Both report the same ZeRO-3 model-loading failure pattern: index-0 out-of-bounds on size-0 tensors during initialization/loading." + "reason": "Both describe the same ZeRO-3 loading bug leading to zero-sized tensors and index-out-of-bounds failures." }, { "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "Different features and failures: kernel loss dispatch vs video chat-template/video-token mismatch." + "reason": "Cross-entropy kernel behavior and video StopIteration are unrelated." }, { "left": "issue:43901", "right": "issue:44512", "accept": false, - "reason": "Both are docs issues, but they reference different API removals and different documentation surfaces." + "reason": "Different documentation mismatches about different APIs/commands." }, { "left": "issue:44841", "right": "issue:45823", "accept": false, - "reason": "Processor loading failure and device-map forward mismatch are distinct issues." + "reason": "Processor loading failure and device_map forward mismatch are different bugs." }, { "left": "issue:45823", "right": "issue:46294", "accept": false, - "reason": "Different models and different failure modes; device placement is not the same as chat-template video token mismatch." + "reason": "Device placement error and chat-template video-token mismatch are different concrete issues." }, { "left": "issue:43975", "right": "issue:46015", "accept": false, - "reason": "Detokenization regression and outdated RAG docs are unrelated." + "reason": "Detokenization bug and docs cleanup are unrelated." }, { "left": "issue:45020", "right": "issue:46304", "accept": false, - "reason": "Remote-code breakage and torch/distributed compatibility are separate regression classes." + "reason": "Remote-code breakage and distributed-build support regression are separate compatibility issues." }, { "left": "issue:43866", "right": "issue:46772", "accept": false, - "reason": "Different model families and different compatibility/load problems." + "reason": "Checkpoint corruption is unrelated to NVFP4 compatibility." }, { "left": "issue:44863", "right": "issue:46015", "accept": false, - "reason": "Model checkpoint loading and docs/examples issues are unrelated." + "reason": "Model checkpoint loading failure vs outdated examples are unrelated." }, { "left": "issue:35308", "right": "issue:42617", "accept": false, - "reason": "IterableDataset multi-GPU crash and 3d_parallel.py failure do not look like the same bug." + "reason": "Both are distributed-training related, but the failure modes and root causes differ." }, { "left": "issue:45812", "right": "issue:45920", "accept": true, - "reason": "Same underlying regression: AutoTokenizer returns wrong token IDs in v5; 45920 is the broader multi-model report and 45812 is a narrower Granite-specific instance." + "reason": "Same AutoTokenizer wrong-token-ID regression; 45920 is just the broader multi-model report." }, { "left": "issue:43872", "right": "issue:46304", "accept": false, - "reason": "bitsandbytes Int8Params incompatibility is unrelated to the distributed-build regression." + "reason": "bitsandbytes kwarg incompatibility is unrelated to the distributed-support issue." }, { - "left": "issue:46901", - "right": "issue:47394", + "left": "issue:39290", + "right": "issue:43531", "accept": false, - "reason": "Both mention double-shifted loss, but they affect different model implementations and the batch-row leak in 47394 makes it a distinct concrete bug." + "reason": "Both mention sliding_window, but one is a missing config field and the other is a different model behavior issue." }, { "left": "issue:44292", "right": "issue:46015", "accept": false, - "reason": "NVFP4 runtime failure and outdated docs are unrelated." + "reason": "NVFP4 runtime failure and docs issues are unrelated." }, { - "left": "issue:39290", - "right": "issue:43531", + "left": "issue:46901", + "right": "issue:47394", "accept": false, - "reason": "Both mention sliding-window concepts, but the model families and concrete failures differ." + "reason": "Similar loss-shift theme, but different models and the second adds cross-row leakage; not clearly the same code path." }, { "left": "issue:43866", "right": "issue:44863", "accept": false, - "reason": "Checkpoint corruption versus inability to load NemotronH checkpoints are not the same underlying issue." + "reason": "Different model-loading failures with different symptoms and likely different causes." }, { "left": "issue:41093", "right": "issue:43638", "accept": false, - "reason": "Both are index/shape errors, but one is a mask-length mismatch and the other is a ZeRO-3 empty-tensor index error." + "reason": "The error strings differ and the reports point to different failure modes, despite both involving ZeRO-3." }, { "left": "issue:43792", "right": "issue:45901", "accept": false, - "reason": "Whisper runtime failure and table-question-answering crash are unrelated." + "reason": "Whisper load failure and table-question-answering crash are unrelated." }, { "left": "issue:43425", "right": "issue:45070", "accept": false, - "reason": "Torch-version incompatibility and a pydantic PretrainedConfig regression are different issues." + "reason": "Torch/transformers version incompatibility and pydantic config-field breakage are different regressions." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "Unable to load a model in tests is not the same as a corrupted checkpoint report." + "reason": "Unit-test load failure and checkpoint corruption are distinct problems." } ] }, "evaluator_result": { "accept": true, - "feedback": "The cluster summary is grounded in the packet and stays conservative. The two accepted soft edges are well-supported by the issue titles: the ZeRO-3 size-0 index errors share the same failure pattern, and 45812/45920 are clearly the same AutoTokenizer v5 wrong-token-ID regression with one being a narrower instance of the broader report. The remaining rejects also read as appropriately cautious, and there were no PR pairs to evaluate." + "feedback": "The summary stays grounded in the packet and the soft-edge judgments are conservative. The two accepted duplicate pairs are well supported by the issue titles: 45812/45920 are both AutoTokenizer wrong-token-ID regressions, and 41762/43596 are both DeepSpeed ZeRO-3 zero-size/index errors. The remaining rejections avoid overstating similarity and mostly distinguish different concrete failure modes or different docs pages." }, "evaluator_used": true, "retried": false, @@ -929,8 +869,8 @@ "issue:43531|issue:45070", "issue:46869|issue:47228", "issue:44387|issue:45005", - "issue:45507|issue:47228", "issue:45020|issue:45800", + "issue:45507|issue:47228", "issue:43716|issue:46294", "issue:44479|issue:46294", "issue:41093|issue:43596", @@ -942,14 +882,14 @@ "issue:39186|issue:44918", "issue:43716|issue:44918" ], - "prepared_review_unit_hash": "9c718b1d3d81d2bba4c7c1e03c55c97c43da4e37c8dc467b741f41b795bd11f4", + "prepared_review_unit_hash": "01f16fb93b739e733167088262eb9746d9f1c19c011e73d14a69dc137c12e8f3", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, @@ -964,265 +904,256 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated import, dtype, quantization, device-map, chat-template, and model-specific regressions. I do not see a single reliable canonical duplicate, and I reject all soft edges.", - "confidence": 0.9, - "canonical_issue_reason": "No clear canonical issue: the items span distinct bugs and error classes across different models and subsystems, so picking one representative would be arbitrary.", + "summary": "This cluster is mostly noisy and cross-cutting: most pairs share only broad keywords like dtype, index, or Qwen/Gemma without the same code path. Only one soft edge looks like a true duplicate-level match: the Qwen3.5 video-input regression pair (44479/46294).", + "confidence": 0.74, + "canonical_issue_reason": "Issue 44479 is the best anchor here because it describes a concrete, reproducible Qwen multimodal regression and has the strongest duplicate-like follow-up in the supplied pairs.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a safe global best because the cluster is not a coherent duplicate set; the titles indicate several unrelated failure modes.", + "best_issue_reason": "Issue 44479 is the most suitable global issue anchor: it is specific, actionable, and closest to a genuine same-bug cluster among the candidates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43541", "right": "issue:47228", "accept": false, - "reason": "Different failures: grouped_mm tracing on Mixtral vs a Sam3 bf16 compilation regression." + "reason": "Different model and failure mode: Mixtral torch-dynamo grouped_mm tracing vs Sam3 bf16 compilation regression." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper runtime failure and a PretrainedConfig/pydantic field break are unrelated bugs." + "reason": "Whisper load failure vs a PretrainedConfig/pydantic compatibility break; unrelated paths." }, { "left": "issue:43596", "right": "issue:45800", "accept": false, - "reason": "DeepSpeed ZeRO-3/BertModel index error is not the same as a torch-transformers version incompatibility." + "reason": "BertModel ZeRO-3 index error vs a broad torch/transformers version incompatibility report." }, { "left": "issue:43975", "right": "issue:45901", "accept": false, - "reason": "Incorrect detokenization for DeepSeek-Coder is unrelated to a table QA task crash." + "reason": "Detokenization bug in v5 vs table-question-answering crash; different subsystems." }, { "left": "issue:43541", "right": "issue:46869", "accept": false, - "reason": "Mixtral torch tracing error and YOLO multi-GPU device-map failure are different code paths." + "reason": "Mixtral compile/tracing issue vs YOLO multi-GPU device-map failure." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Both involve Qwen3.5, but one is a generic chat crash and the other is a streaming chat/completions failure; not enough to treat as the same bug." + "reason": "Same general Qwen3.5 family, but different stacks and paths: Transformers chat vs vLLM streaming API." }, { "left": "issue:44292", "right": "issue:45901", "accept": false, - "reason": "Qwen-3 NVFP4 loading/runtime error and table QA crash are unrelated." + "reason": "Qwen NVFP4 load failure vs table QA crash; no shared bug." }, { "left": "issue:39290", "right": "issue:45823", "accept": false, - "reason": "Gemma3 sliding_window_pattern missing and Gemma4 device mismatch are distinct issues." + "reason": "Gemma3 sliding-window config error vs Gemma4 device placement mismatch." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "Importing GenerationMixin and a Gemma3 config attribute error are unrelated." + "reason": "Missing GenerationMixin import and Gemma3 sliding_window_pattern are unrelated issues." }, { "left": "issue:45464", "right": "issue:45685", "accept": false, - "reason": "Qwen streaming API failure and an MPS histogram error on MoE are different bugs." + "reason": "Qwen3.5 streaming inference failure vs MPS histogram op limitation; different code paths." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Both mention MoE, but the traced grouped_mm error and the MPS histogram failure are not the same code-path problem." + "reason": "Mixtral compile bug vs MPS int histogram limitation." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Grouped_mm tracing in Mixtral and autocast dtype mismatch in Phi-tiny-MoE are separate failures." + "reason": "Torch-dynamo grouped_mm failure vs autocast dtype mismatch; not the same bug." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Processor failure for Voxtral and a compile error for non-template nodes are unrelated." + "reason": "Voxtral processor failure vs template compilation error; unrelated." }, { "left": "issue:45507", "right": "issue:46869", "accept": false, - "reason": "GraniteMoEHybrid invalid-method call and YOLO device-map failure do not describe the same bug." + "reason": "GraniteMoE invalid method vs YOLO multi-GPU device-map issue." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "Tied lm_head weights in Mistral3 and a corrupted Ovis2 checkpoint are not the same issue." + "reason": "Tied lm_head bug in Mistral3 vs corrupted Ovis2 checkpoint." }, { "left": "issue:45399", "right": "issue:46661", "accept": false, - "reason": "Flash-attn fallback blocking and remote-code model_class/config_class errors are different problems." + "reason": "flash-attn2 fallback gating vs remote-code model config_class error." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig is not the same as the pydantic PretrainedConfig field regression." + "reason": "StableLm pad_token_idx regression vs generic PretrainedConfig field break; too broad to merge." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Both are IndexError reports, but one is a mask/tensor shape mismatch and the other is an out-of-bounds index during ZeRO-3 loading." + "reason": "Both are index/shape errors, but one is mask-length mismatch and the other is DeepSpeed ZeRO-3 empty-load failure." }, { "left": "issue:45800", "right": "issue:45823", "accept": false, - "reason": "Torch/transformers version incompatibility and Gemma4 device mismatch are unrelated." + "reason": "Torch/transformers incompatibility vs Gemma4 device-map mismatch; not the same underlying bug." }, { "left": "issue:43872", "right": "issue:47228", "accept": false, - "reason": "bitsandbytes Int8Params constructor mismatch is unrelated to the Sam3 compilation regression." + "reason": "bitsandbytes constructor incompatibility vs Sam3 bf16 compilation regression." }, { "left": "issue:44805", "right": "issue:44918", "accept": false, - "reason": "Mask shape mismatch and Qwen3.5 embedding unpacking with TRL SFT are different failure modes." + "reason": "Similar shape-error wording, but one is a mask/index mismatch and the other is TRL embedding unpacking." }, { "left": "issue:43716", "right": "issue:47228", "accept": false, - "reason": "Mistral-3 image preprocessor dtype mismatch is unrelated to Sam3 bf16 compilation regression." + "reason": "Mistral image-preprocessor dtype mismatch vs Sam3 compilation regression." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE is not the same as Qwen3.5 streaming API failure." + "reason": "Autocast dtype mismatch in Phi-tiny-MoE vs Qwen3.5 streaming API failure." }, { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "PaddleOCR-VL load failure and GGUF gpt-oss support are different model support issues." + "reason": "PaddleOCR-VL load problem in vLLM vs GGUF gpt-oss architecture support." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "GGUF architecture support and Qwen chat streaming failure are unrelated." + "reason": "GGUF architecture support vs Qwen chat/streaming failure; unrelated." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "init_empty_weights/_is_hf_initialized TypeError is a different regression from the pydantic PretrainedConfig field break." + "reason": "init_empty_weights/_is_hf_initialized constructor issue vs PretrainedConfig pydantic regression." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and a PretrainedConfig pydantic break are not the same issue." + "reason": "Qwen3-MoE sliding_window bug vs generic config-field breakage." }, { "left": "issue:46869", "right": "issue:47228", "accept": false, - "reason": "YOLO multi-GPU device-map failure and Sam3 bf16 compilation regression are unrelated." + "reason": "YOLO device-map problem vs Sam3 bf16 compilation regression." }, { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "int4 CUDA memory growth and tied-weights translation-model issues are different regressions." + "reason": "int4 memory/OOM regression vs tied-weights translation-model bug." }, { - "left": "issue:45507", - "right": "issue:47228", + "left": "issue:45020", + "right": "issue:45800", "accept": false, - "reason": "GraniteMoEHybrid invalid method and Sam3 compilation regression are unrelated." + "reason": "Remote-code breakage and torch/transformers incompatibility are both broad version issues, but not the same concrete bug." }, { - "left": "issue:45020", - "right": "issue:45800", + "left": "issue:45507", + "right": "issue:47228", "accept": false, - "reason": "A broad remote_code regression and a specific torch 2.4.1/transformers 5.8.0 incompatibility are not the same concrete bug." + "reason": "GraniteMoE invalid method vs Sam3 compilation regression." }, { "left": "issue:43716", "right": "issue:46294", "accept": false, - "reason": "Image preprocessor dtype mismatch and Qwen3.5 video token/feature mismatch are different multimodal issues." + "reason": "Both are multimodal input mismatches, but image-preprocessor dtype failure and video token/feature mismatch are different bugs." }, { "left": "issue:44479", "right": "issue:46294", - "accept": false, - "reason": "Both are video-input regressions, but they affect different model families and describe different symptoms, so this is too broad to merge." - }, - { - "left": "issue:41093", - "right": "issue:43596", - "accept": false, - "reason": "The mask-shape IndexError and the ZeRO-3/BertModel out-of-bounds IndexError are not clearly the same bug." - }, - { - "left": "issue:43054", - "right": "issue:43994", - "accept": false, - "reason": "Both concern SigLIP2 quality, but one reports worse embeddings and the other nonsensical AutoModel/pipeline outputs; not enough evidence of one concrete bug." + "accept": true, + "reason": "Same Qwen video-input regression family with the same apply_chat_template/video-token mismatch theme; plausible same bug." }, { "left": "issue:46468", "right": "issue:46869", "accept": false, - "reason": "Broken official Nemotron example and YOLO multi-GPU device-map failure are unrelated." + "reason": "Nemotron example failure vs YOLO device-map issue; unrelated." }, { "left": "issue:45759", "right": "issue:45902", "accept": false, - "reason": "Qwen 3.5 composite config unwrap behavior and an uninitialized-memory rotary embedding bug are different issues." + "reason": "Composite-config unwrapping bug vs uninitialized-memory rotary embedding bug in Qwen3.5 MoE." }, { "left": "issue:46640", "right": "issue:46661", "accept": false, - "reason": "Flash Attention 2 issue and remote-code config_class AttributeError are unrelated." + "reason": "Flash Attention 2 issue vs remote-code model loading error." }, { "left": "issue:43596", "right": "issue:46165", "accept": false, - "reason": "ZeRO-3 loading failure for BertModel is not the same as resuming training with locally downloaded remote-code models." + "reason": "ZeRO-3 BertModel load failure vs resume-with-remote-code training issue." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP weight dimensionality error and Qwen3.5 embedding unpacking failure are unrelated." + "reason": "FSDP 2-D weight error vs TRL embedding unpacking error." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Image dtype mismatch and TRL SFT embedding unpacking are different code paths." + "reason": "Mistral multimodal dtype issue vs Qwen3.5 SFT unpacking issue." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The summary stays within the packet, and the only accepted soft edge (44479/46294) is plausibly the same Qwen video-input regression family based on the titles. The rejections also avoid overclaiming shared bugs from broad keyword overlap." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -1321,226 +1252,226 @@ "issue:45987|issue:46858", "issue:43950|issue:45902" ], - "prepared_review_unit_hash": "67a8b29d55b669c9049a1517fec990d39fc397c5eac48380b3c226360c4e5986", + "prepared_review_unit_hash": "6cf6cbe577800ab6e6cdf579e4bff21e4fd6900503b160778ed04b4058616a6f", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37137, - "estimated_input_tokens": 9285, - "estimated_eval_tokens": 18826 + "serialized_chars": 37135, + "estimated_input_tokens": 9284, + "estimated_eval_tokens": 18824 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous, with a few true duplicate subclusters: non-persistent buffer corruption in v5, cache length return-type regressions, Gemma4 quantized embedder dtype casting, and offloading/disk-save issues. Most other soft pairs are false positives that only share a broad subsystem or model family.", - "confidence": 0.86, - "canonical_issue_reason": "No single canonical issue cleanly covers the whole set because the items split into several unrelated bug families; if forced to pick one representative, #43950 is the strongest broad regression with clear duplicate support.", + "summary": "This set is mostly a mix of unrelated issues, with a few clear duplicate families around non-persistent buffer initialization, cache sequence-length return types, offloading save failures, and a couple of model-specific regressions.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43644 is the clearest root-cause report in the non-persistent-buffer regression family: it states the broad bug plainly and matches nearby duplicates like 43950/44534.", "canonical_pr_reason": null, - "best_issue_reason": "#43950 is the best issue candidate overall because it is a clear, well-scoped regression and it has close duplicate overlap with #43644 and #44534.", + "best_issue_reason": "issue:43644 is the strongest single representative issue overall: broad, specific, and already has close duplicate reports that point to the same regression.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Unrelated regressions: custom-model notebook initialization vs flash-attn2 fallback gating." + "reason": "Different failures: notebook custom-model init vs flash-attn2 fallback gating." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "Different models and failures: GPT-OSS perplexity issue vs GLM5 problem." + "reason": "Different models and symptoms; both are perplexity-related but not the same bug." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Qwen3Moe FP8 config failure is unrelated to flash-attn2 fallback checks." + "reason": "Qwen3Moe FP8 failure is unrelated to flash-attn2 fallback checks." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Both involve tied weights, but the model families and bugs differ; not the same concrete issue." + "reason": "Different tied-weight issues in different model families; not the same concrete bug." }, { "left": "issue:45632", "right": "issue:45698", - "accept": false, - "reason": "Both touch custom remote-code loading, but one is a cache-path collision and the other is wrong module selection after save/load." + "accept": true, + "reason": "Both describe the same remote-code/custom-module path collision class causing the wrong module to load." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization/offloading behavior is unrelated to tied-weights regression." + "reason": "Quantization/offloading behavior is unrelated to tied-weights regressions." }, { "left": "issue:35308", "right": "issue:36331", "accept": false, - "reason": "IterableDataset multi-GPU crash and compute_loss signature change are distinct problems." + "reason": "Different problems: iterable-dataset multi-GPU crash vs CustomTrainer compute_loss signature mismatch." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "DSA indexer behavior is unrelated to a removed docs command reference." + "reason": "Unrelated DSA/indexer discussion vs docs cleanup." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "A cache-strategy typo and a docs command removal are not the same bug." + "reason": "A cache-strategy typo is unrelated to the removed CLI command doc issue." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "TRL SFT embedding unpacking and GraniteMoEHybrid invalid method are different code paths." + "reason": "Different model/trainer failures with no clear shared code path." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache/linear_attention error is unrelated to tied-weight handling." + "reason": "PagedAttentionCache crash is unrelated to tied-weight regressions." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Different tokenizer/runtime failures on different model families; not one concrete issue." + "reason": "Different tokenizer/runtime problems on different models." }, { "left": "issue:45183", "right": "issue:46463", - "accept": true, - "reason": "46463 is a specific manifestation of the broader input_ids/inputs_embeds XOR validation/error-message bug." + "accept": false, + "reason": "Both touch input validation, but the titles point to different concrete bugs and scopes." }, { "left": "issue:35308", "right": "issue:42915", "accept": false, - "reason": "Training crash with IterableDataset is unrelated to Qwen3Moe FP8 failure." + "reason": "Multi-GPU dataset crash is unrelated to Qwen3Moe FP8 issues." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading and translation-model tied weights are unrelated." + "reason": "Model-loading failure is unrelated to tied-weights regressions." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "Docs example mistakes are distinct from a model-quality/embedding regression." + "reason": "Docs/example errors vs model quality regression are different issues." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Both describe the same non-persistent-buffer corruption/junk-fill regression in transformers v5." + "reason": "Same non-persistent-buffer corruption regression, just reworded." }, { "left": "issue:46489", "right": "issue:46491", "accept": false, - "reason": "Tokenizer regressions on different model families; too model-specific to treat as one bug." + "reason": "Different tokenizer bugs affecting different models/symptoms." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Unexpected compute_loss kwarg and ZeRO-3 loading IndexError are unrelated." + "reason": "Unexpected compute_loss kwarg vs ZeRO-3 loading crash are different bugs." }, { "left": "issue:43644", "right": "issue:43950", "accept": true, - "reason": "Same underlying v5 regression: non-persistent buffers get corrupted or filled with junk during load/init." + "reason": "Same regression: non-persistent buffers get corrupted/junk-filled after load/materialization." }, { "left": "issue:33666", "right": "issue:35308", "accept": false, - "reason": "Qwen2-VL multi-GPU training is not the same bug as generic IterableDataset length mismatch crashes." + "reason": "Multi-GPU training for Qwen2-VL is not the same as the iterable-dataset crash." }, { "left": "issue:46032", "right": "issue:47246", "accept": true, - "reason": "Both point to the same Mamba2 inter-chunk recurrence bug in the slow-path implementation, with NemotronH missing the propagated fix." + "reason": "Both describe the same Mamba2 inter-chunk recurrence bug, with NemotronH as a missing propagation of the same fix." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Different bitsandbytes/storage compatibility failures with different error surfaces." + "reason": "Different bitsandbytes/storage incompatibilities with different failure modes." }, { "left": "issue:46899", "right": "issue:46932", "accept": true, - "reason": "46932 explicitly refers to the same Gemma4 quantized vision/audio embedder dtype-casting bug." + "reason": "Same Gemma4 quantized vision/audio embedder dtype-casting bug; 46932 is an explicit follow-up/dup of the same fix." }, { "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "SigLIP2 embedding quality regression and tokenizer-version changes are unrelated." + "reason": "Different tokenizer-related complaints; not the same concrete bug." }, { "left": "issue:45920", "right": "issue:46489", "accept": false, - "reason": "Broad AutoTokenizer wrong-ID regression and DeepSeek-Coder v1 detokenization bug are distinct issues." + "reason": "General AutoTokenizer regression vs a specific DeepSeek-Coder tokenizer bug." }, { "left": "issue:47332", "right": "issue:47333", "accept": true, - "reason": "Near-identical offloading/disk-save failures for Qwen3 VL/Qwen3-A3B models; same underlying issue." + "reason": "Same offloading save failure for disk-offloaded Qwen3 models, just on different model variants." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Both involve SigLIP2, but zero probabilities and poorer text embeddings are not clearly the same concrete defect." + "reason": "Both mention SigLIP2, but one is zero probabilities and the other is degraded embeddings; not the same bug." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Autocast dtype mismatch and Qwen3.5 RLHF NaN explosion have different root causes." + "reason": "Torch autocast dtype mismatch and RLHF NaN from position_ids are different failure mechanisms." }, { "left": "issue:45987", "right": "issue:46858", "accept": true, - "reason": "Same cache API regression: get_seq_length() returns the wrong kind of tensor instead of a plain int-like value." + "reason": "Same cache API contract issue: get_seq_length returns the wrong tensor form instead of an int." }, { "left": "issue:43950", "right": "issue:45902", - "accept": false, - "reason": "Non-persistent buffer corruption is unrelated to Qwen3_5Moe rotary inv_freq uninitialized-memory access." + "accept": true, + "reason": "Same non-persistent-buffer initialization regression; 45902 is a model-specific manifestation." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays appropriately conservative. The representative issue choice (#43950) is supported by the duplicate-like overlap with #43644 and #44534, and the cluster-level note that most other pairs are false positives matches the packet. The accepted soft pairs also look defensible from the titles and available metadata. No obvious overstatement or mergeability issue is present." + "feedback": "The analyst output is generally grounded in the packet and the soft-edge verdicts stay conservative. The accepted pairs mostly have clear title-level overlap in the same bug family (non-persistent buffers, cache length contract, Gemma4 dtype casting, offloading save failures, and related custom-module loading issues). I don\u2019t see a clear overstatement that would warrant rejection." }, "evaluator_used": true, "retried": false, @@ -1614,8 +1545,8 @@ "issue:43054|issue:43493", "issue:44910|issue:45859", "issue:45663|issue:45902", - "issue:42947|issue:44387", "issue:44403|issue:44451", + "issue:42947|issue:44387", "issue:24643|issue:28530", "issue:43975|issue:45237", "issue:41762|issue:42915", @@ -1627,13 +1558,13 @@ "issue:45081|issue:45399", "issue:43824|issue:43872", "issue:45020|issue:45823", + "issue:44661|issue:45020", "issue:43782|issue:44960", "issue:41553|issue:43883", - "issue:44661|issue:45020", "issue:43828|issue:45823", "issue:45020|issue:45356", - "issue:42491|issue:44368", "issue:43872|issue:46869", + "issue:42491|issue:44368", "issue:41553|issue:43881", "issue:45759|issue:46661", "issue:45020|issue:46661", @@ -1643,229 +1574,229 @@ "issue:43827|issue:44509", "issue:43881|issue:43883" ], - "prepared_review_unit_hash": "a3d9bc108d04c0e5243544a3bb268ae46d6064a982207d953a61b0d6e7a01b0b", + "prepared_review_unit_hash": "bc878288203857719e234ff00567e5a464f13f9d3e690c572e52f8c6663a3562", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37489, - "estimated_input_tokens": 9373, - "estimated_eval_tokens": 19002 + "serialized_chars": 37488, + "estimated_input_tokens": 9372, + "estimated_eval_tokens": 19000 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous and centered on loading/tokenizer/regression reports. The only clearly supported duplicate pair is the v5 docs pipeline-removal reports (43827/44509); the SigLIP2 pair is related in theme but not established as the same bug.", - "confidence": 0.78, - "canonical_issue_reason": null, + "summary": "This cluster is mostly a loose collection of unrelated Transformers regressions around model loading, tokenizers, remote-code configs, MoE/LoRA training, and a couple of docs issues. Only the two v5 pipeline-doc tickets look like the same underlying change; the rest are too model- or code-path-specific to merge as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 45020 is the broadest representative: it covers the general class of recent v5 remote-code/model-loading breakages that several other issues touch, even though this cluster is not a single duplicate family.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45020 is the broadest representative of the cluster\u2019s repeated remote-code/loading regression theme, but the set is mixed enough that it should not be treated as a true canonical duplicate.", + "best_issue_reason": "Issue 45020 is the best overall issue to anchor the cluster because it is the clearest umbrella report, has strong discussion activity, and aligns with the dominant theme of v5 loading regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43054", "right": "issue:43493", "accept": false, - "reason": "Both concern SigLIP2 quality, but one is a specific text-embedding regression report and the other is a broader HF-vs-JAX discrepancy; the packet doesn\u2019t show they are the same bug." + "reason": "Both are SigLIP2-related, but one is a quality/regression report and the other is an HF-vs-JAX implementation discrepancy; different bugs and fixes." }, { "left": "issue:44910", "right": "issue:45859", "accept": false, - "reason": "Different failure modes: flash-attn packed-sequence crash vs CPU-offload rotary embedding incompatibility." + "reason": "One is a flash-attention crash from 3D position_ids handling; the other is CPU-offload incompatibility in rotary embedding. Related model family, different code paths." }, { "left": "issue:45663", "right": "issue:45902", "accept": false, - "reason": "Gemma4 FSDP2 shared_kv_states bug is unrelated to Qwen3_5MoeTextRotaryEmbedding uninitialized memory." + "reason": "Gemma4 FSDP2 shared_kv_states bug vs Qwen3_5MoE inv_freq materialization bug; completely different models and failure modes." }, { - "left": "issue:42947", - "right": "issue:44387", + "left": "issue:44403", + "right": "issue:44451", "accept": false, - "reason": "One is LoRA gradient-checkpointing ineffectiveness; the other is an int4 quantization memory regression." + "reason": "Generic load-noise complaint vs a specific model load failure for ScandiBERT; not the same underlying bug." }, { - "left": "issue:44403", - "right": "issue:44451", + "left": "issue:42947", + "right": "issue:44387", "accept": false, - "reason": "Generic load-noise report is not the same as a specific ScandiBERT load failure." + "reason": "Gradient checkpointing with PEFT LoRA is a training-behavior issue; the other is int4 quantization causing higher reserved memory. Both memory-adjacent, but not the same bug." }, { "left": "issue:24643", "right": "issue:28530", "accept": false, - "reason": "DeepSpeed training weight-shape error and early-stopping metric configuration are unrelated." + "reason": "DeepSpeed weight-shape RuntimeError and early-stopping metric selection are unrelated training issues." }, { "left": "issue:43975", "right": "issue:45237", "accept": false, - "reason": "Tokenizer detokenization bug is unrelated to AMD GPU execution failure." + "reason": "Tokenizer detokenization regression vs AMD GPU support failure; no shared code path." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Gemma3 ZeRO-3 index error and Qwen3Moe FP8 failure are different model-path bugs." + "reason": "Gemma3 ZeRO-3 load failure vs Qwen3-MoE FineGrainedFP8Config failure; both loading-related, but distinct model/config bugs." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "Both are loading complaints, but not the same concrete bug or model path." + "reason": "Both mention loading failures, but one is a generic transformer noise report and the other is a specific sleng-bert model load breakage." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "Remote-code config field loss is not clearly the same concrete issue as broad recent remote_code breakage." + "reason": "AutoConfig losing fields with trust_remote_code is a config-construction bug, not the broader remote-code loading regressions described in 45020." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer loading regression and flash-attn2 fallback gating are unrelated." + "reason": "Tokenizer loading failure for est-roberta vs flash-attn fallback blocking; unrelated subsystems." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "LoRA checkpointing inefficiency and Qwen3 MoE memory usage are different training regressions." + "reason": "Both involve training memory/perf, but one is ineffective gradient checkpointing with LoRA and the other is inefficient Qwen3 MoE memory usage." }, { "left": "issue:45685", "right": "issue:45823", "accept": false, - "reason": "MPS histogram/int error is unrelated to Gemma4 device-map mismatch." + "reason": "MPS histogram op not implemented for Int vs Gemma4 device-mismatch under device_map=auto; different platform bugs." }, { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Mistral regex patch crash and flash-attn2 fallback checks are different code paths." + "reason": "Mistral regex patch crash in tokenizer loading vs flash-attn fallback gating; no common underlying change." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Qwen2.5-VL import error and bitsandbytes Int8Params compatibility are unrelated." + "reason": "Missing Qwen2.5-VL import and bitsandbytes Int8Params incompatibility are unrelated breakages." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Generic remote_code breakage is not the same as Gemma4 PLE device placement." + "reason": "Remote-code model breakage and Gemma4 device-mapping mismatch are different issues." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Tokenizer-mapping-name failure in add-new-model-like is a registry bug, not the broader remote-code regression." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Qwen3VL weight_only error and GLM5 loading are different issues." + "reason": "Qwen3VL weight_only load error and GLM5 are different model-specific failures." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Bad AutoTokenizer message for Voxtral is unrelated to Molmo tied-weights attribute failure." - }, - { - "left": "issue:44661", - "right": "issue:45020", - "accept": false, - "reason": "add-new-model-like mapping failure is not the same bug as remote_code loading regressions." + "reason": "Bad AutoTokenizer error message for Voxtral vs missing all_tied_weights_keys in Molmo; unrelated bugs." }, { "left": "issue:43828", "right": "issue:45823", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to Gemma4 device-map mismatch." + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is a computation issue, while Gemma4 PLE has a device placement mismatch." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "The remote-code theme overlaps, but the titles describe different concrete failures and should not be merged." + "reason": "Remote-code breakage vs Kimi-K2.5 tokenizer codec regression; different symptoms and fixes." }, { - "left": "issue:42491", - "right": "issue:44368", + "left": "issue:43872", + "right": "issue:46869", "accept": false, - "reason": "LoRA hf4.x/hf5.x incompatibility is unrelated to the tie_word_embeddings warning during Qwen3.5 fine-tuning." + "reason": "bitsandbytes parameter init incompatibility is unrelated to YOLO multi-GPU device-map failure." }, { - "left": "issue:43872", - "right": "issue:46869", + "left": "issue:42491", + "right": "issue:44368", "accept": false, - "reason": "bitsandbytes Int8Params incompatibility is unrelated to YOLO multi-GPU device-map failure." + "reason": "LoRA Qwen3-MoE incompatibility on hf4.x vs a tie_word_embeddings warning in ms-swift; same family, but not the same bug." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Voxtral tokenizer error message and glm-4v-9b load failure are different problems." + "reason": "Voxtral tokenizer error-message issue vs GLM-4V load failure; different models and failure paths." }, { "left": "issue:45759", "right": "issue:46661", "accept": false, - "reason": "Composite config unwrapping bug is not the same as remote-code config_class attribute failure." + "reason": "AutoModel.from_config not unwrapping text_config vs remote-code models missing config_class; both model-construction issues, but not the same root cause." }, { "left": "issue:45020", "right": "issue:46661", "accept": false, - "reason": "The remote-code theme overlaps, but the titles describe different concrete failures and should not be merged." + "reason": "General remote-code regressions and a specific config_class AttributeError are related only at a very high level." }, { "left": "issue:45488", "right": "issue:46489", "accept": false, - "reason": "Both concern tokenizer regressions, but they affect different models and different underlying tokenization paths." + "reason": "DeepSeek tokenizer pre-tokenizer override vs DeepSeek-Coder v1 detokenization regression; related model family, different tokenizer bugs." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "Mistral3 tied-weights bug is unrelated to non-persistent buffer corruption in from_pretrained." + "reason": "Mistral3 lm_head tying bug vs non-persistent buffer corruption in from_pretrained; unrelated code paths." }, { "left": "issue:45715", "right": "issue:46315", "accept": false, - "reason": "Different tokenizer performance regressions in opposite conversion methods; not the same bug." + "reason": "Both are tokenizer performance regressions, but one is convert_ids_to_tokens and the other is convert_tokens_to_ids; not mergeable as one bug." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Both report the same v5 docs problem: removed pipeline tasks still referenced in documentation." + "reason": "These are the same docs regression: outdated pipeline-task docs still reference summarization/translation/text2text-generation after v5 removals." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "GLM-4V-9B load failure and Molmo tied-weights attribute error are distinct model issues." + "reason": "GLM-4V load failure and Molmo tied-weights attribute error are different model-specific issues." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The summary is mostly grounded, but one soft-edge acceptance overstates the evidence. The v5 docs pair (43827/44509) is well supported, but 43054/43493 is only loosely similar from the titles: one is a quality regression report for SigLIP2 text embeddings, the other is a broader HF-vs-JAX implementation discrepancy. The packet does not establish that these are the same bug, so this should be treated more conservatively as a possible related issue, not a true duplicate." + "accept": true, + "feedback": "The summary is grounded in the packet, and the one accepted soft pair (43827/44509) looks mergeable as the same docs regression. The other soft-edge rejections are generally conservative. Minor wording note: 'strong discussion activity' for 45020 is a bit stronger than the packet evidence strictly shows, but it does not materially distort the cluster assessment." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1933,8 +1864,8 @@ "issue:46897" ], "soft_pairs": [ - "issue:43792|issue:44220", "issue:43257|issue:45305", + "issue:43792|issue:44220", "issue:36331|issue:41093", "issue:46894|issue:46897", "issue:42947|issue:45923", @@ -1953,9 +1884,9 @@ "issue:45325|issue:45902", "issue:43366|issue:44589", "issue:43638|issue:46515", - "issue:41553|issue:44991", "issue:43716|issue:44560", "issue:45663|issue:45823", + "issue:41553|issue:44991", "issue:42915|issue:43957", "issue:44155|issue:45584", "issue:43824|issue:43957", @@ -1964,208 +1895,220 @@ "issue:43950|issue:45834", "issue:44560|issue:44918" ], - "prepared_review_unit_hash": "f5821c080dbb56aa28ef82f409e0566d9690b9a9a907d902ea7c7840151b24a3", + "prepared_review_unit_hash": "9133502e6ac13e37c9879c6ead242bc7481bcfb52f1ecae988d9d31d7496a5d6", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36760, - "estimated_input_tokens": 9190, - "estimated_eval_tokens": 18636 + "serialized_chars": 36767, + "estimated_input_tokens": 9192, + "estimated_eval_tokens": 18640 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated reports with a couple of likely duplicate pairs: the Qwen VL video-input regression pair and the training-loss double-shift pair. The rest are best treated as separate issues.", - "confidence": 0.68, - "canonical_issue_reason": "Issue 46894 is the clearest standalone report of the underlying training-loss double-shift bug and is more representative than the narrower model-specific follow-up.", + "summary": "This set is heterogeneous overall, but there are a few plausible duplicate pairs: Qwen ZeRO-3 weight-loading regressions, Qwen/VL video-input regressions, a repeated mask-shape IndexError, and a repeated training-loss double-shift bug. Most other pairs are clearly unrelated despite superficial subsystem overlap.", + "confidence": 0.66, + "canonical_issue_reason": "No single canonical issue fits the whole cluster because the items split into several unrelated bug groups. If one umbrella issue is needed for the strongest subgroup, issue 44479 is the best representative because it describes a concrete 5.3.0 video-input regression across multiple Qwen/VL variants.", "canonical_pr_reason": null, - "best_issue_reason": "46894 is a strong anchor issue: explicit bug pattern, clear scope, and easy to relate to follow-up reports like 46897.", + "best_issue_reason": "Issue 44479 is the best global issue for its subgroup: it states the version-specific regression clearly, spans multiple affected models, and is more umbrella-like than the narrower model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43792", - "right": "issue:44220", + "left": "issue:43257", + "right": "issue:45305", "accept": false, - "reason": "Both are audio/Whisper-related, but one is a broad runtime failure and the other is a specific fbank helper issue; not clearly the same bug." + "reason": "Both mention DeepSpeed and Qwen, but one is a weight-conversion failure and the other is a GAS gradient-averaging bug; different code paths." }, { - "left": "issue:43257", - "right": "issue:45305", + "left": "issue:43792", + "right": "issue:44220", "accept": false, - "reason": "Both mention DeepSpeed, but one is missing Qwen3 MoE weight conversion and the other is gradient averaging with GAS/model_accepts_loss_kwargs; different failures." + "reason": "Both are audio/Whisper-adjacent, but the titles point to different failure modes and there is no clear shared code-path bug from the available evidence." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Unexpected compute_loss kwarg vs mask/tensor shape mismatch are different code paths and symptoms." + "reason": "One is a Trainer API kwarg breakage; the other is a tensor mask/index shape mismatch. These are unrelated defects." }, { "left": "issue:46894", "right": "issue:46897", "accept": true, - "reason": "Same double-shift training-loss bug described for two model families; these look like duplicate reports of one underlying trainer issue." + "reason": "Both describe the same training-loss double-shift pattern across different models, suggesting one shared loss-computation fix could cover both." }, { "left": "issue:42947", "right": "issue:45923", "accept": false, - "reason": "Both touch gradient checkpointing, but one is ineffective LoRA checkpointing behavior and the other is a missing capability flag." + "reason": "Gradient-checkpointing ineffective with LoRA is not the same as a missing supports_gradient_checkpointing flag in a trust_remote_code model." }, { "left": "issue:43940", "right": "issue:45313", - "accept": false, - "reason": "Both are ZeRO-3 weight-loading failures on Qwen variants, but the model targets and missing-weight patterns differ enough to treat separately." + "accept": true, + "reason": "Both are Qwen-family DeepSpeed ZeRO-3 weight-loading failures with missing parameters, so they plausibly share the same loading-path bug." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "DeepSpeed training RuntimeError and void segmentation-map processing are unrelated." + "reason": "DeepSpeed training weight shape error and void segmentation-map preprocessing are unrelated." }, { "left": "issue:44479", "right": "issue:44560", "accept": true, - "reason": "Both describe the same 5.3.0 Qwen VL video-input regression; 44560 appears to be a specific StopIteration manifestation of the broader bug in 44479." + "reason": "Both are Qwen/VL video-input regressions in transformers 5.3.0; 44560 looks like a specific manifestation of the broader video bug in 44479." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Different model-loading failures for different architectures; no shared code-path bug is evident." + "reason": "GLM-4.7 unit-test loading failure and NemotronH checkpoint loading failure are separate model-loading issues." }, { "left": "issue:41093", "right": "issue:44805", - "accept": false, - "reason": "Both are generic index/mask shape errors, but there is no evidence they come from the same underlying model or code path." + "accept": true, + "reason": "The titles describe the same mask-vs-indexed-tensor shape mismatch error; the numeric sizes differ but the underlying bug appears identical." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "GGUF architecture support and autocast dtype mismatch are unrelated." + "reason": "GGUF gpt-oss support and a torch.autocast dtype mismatch on Phi-tiny-MoE are unrelated." }, { "left": "issue:43716", "right": "issue:46468", "accept": false, - "reason": "Mistral-3 image-preprocessor dtype mismatch and Nemotron example breakage are unrelated." + "reason": "An image-preprocessor dtype mismatch for Mistral-3 is not the same as a broken Nemotron example." }, { "left": "issue:44928", "right": "issue:45588", "accept": false, - "reason": "RLHF NaN/SDPA math fallback and flash-attention s_aux=None crash are different bugs." + "reason": "One is a Qwen3.5 RLHF NaN/SDPA fallback issue; the other is a flash-attention AttributeError on sink-less models. Different failures and code paths." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "GGUF gpt-oss support and Qwen3-MoE sliding_window behavior are unrelated." + "reason": "GGUF architecture support and Qwen3-MoE sliding-window behavior are unrelated." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "PagedAttentionCache linear_attention crash and NemotronH checkpoint loading are different issues." + "reason": "PagedAttentionCache linear_attention crash and NemotronH checkpoint loading are different bugs." }, { "left": "issue:43596", "right": "issue:46515", "accept": false, - "reason": "Zero-3/Bert index error and torch version incompatibility are not the same bug." + "reason": "Zero-3/BertModel out-of-bounds indexing and torch-version incompatibility are not the same defect." }, { "left": "issue:44863", "right": "issue:46402", "accept": false, - "reason": "Both are load failures, but for different models and different failure modes." + "reason": "Both are load failures, but for different model families and different failure mechanisms." }, { "left": "issue:45325", "right": "issue:45902", "accept": false, - "reason": "Different rotary/position-id bugs with different root causes: temporal scaling vs uninitialized inv_freq after materialization." + "reason": "One is a Qwen2.5-VL rope-scaling bug; the other is uninitialized memory after meta/to_empty materialization. Different underlying problems." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "GGUF support and missing Float8 storage are unrelated." + "reason": "gpt-oss GGUF support and missing Float8 storage are unrelated compatibility issues." }, { "left": "issue:43638", "right": "issue:46515", "accept": false, - "reason": "Deepspeed ZeRO-3 init failure and torch 2.4.1 incompatibility are different issues." - }, - { - "left": "issue:41553", - "right": "issue:44991", - "accept": false, - "reason": "Both are tokenizer-loading complaints, but one is a bad error message for Voxtral and the other is an actual tokenizer incompatibility for a different model." + "reason": "A ZeRO-3 init/indexing failure with Bert and a torch version incompatibility are different issues." }, { "left": "issue:43716", "right": "issue:44560", "accept": false, - "reason": "Image-preprocessor dtype mismatch and Qwen3-vl video StopIteration are unrelated." + "reason": "Image-preprocessor dtype mismatch and Qwen3-VL video StopIteration are both multimodal, but they are not the same bug." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Both are Gemma4-related, but one is an FSDP2 attention KeyError and the other is a device-map forward mismatch." + "reason": "One is a per-layer shared_kv_states KeyError in training; the other is a device-mapping mismatch in forward. Not the same code-path problem." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "A bad tokenizer error message for Voxtral is not the same as an actual tokenizer load failure for EMBEDDIA/est-roberta." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "FineGrainedFP8 failure on Qwen3Moe and meta-device loading regression are different problems." + "reason": "Qwen3Moe FP8 failure and meta-device loading breakage are separate model/runtime incompatibilities." }, { "left": "issue:44155", "right": "issue:45584", "accept": false, - "reason": "AudioFlamingo track-leak batching bug and Whisper empty-transcription failure are unrelated audio issues." + "reason": "AudioFlamingo3 track-leak inference corruption and Whisper empty-transcription failure after special-token alignment are different bugs." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Missing Qwen2_5_VL import and meta-device loading regression are different failure modes." + "reason": "Missing Qwen2_5_VL import and meta-device loading breakage are unrelated." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "3d_parallel example failure and GGUF gpt-oss support are unrelated." + "reason": "3d_parallel.py runtime failure and GGUF gpt-oss support are unrelated." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Qwen video StopIteration and an MPS histogram Int limitation are different hardware/model bugs." + }, + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Non-persistent buffer corruption in from_pretrained and a long-OCR index error are unrelated." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Qwen video StopIteration and TRL SFT embedding unpacking are different code paths." + "reason": "Video-loading StopIteration and SFT trainer embedding unpacking failure are different failure modes." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The two accepted duplicate pairs are plausibly same-bug reports based on the titles/pattern references, and the rest are appropriately treated as separate issues. No clear overstatement of evidence." + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The identified duplicate pairs are plausible from the titles/body signals, and the claim that the cluster is heterogeneous is well supported. The umbrella selection of issue 44479 is also reasonable because it is explicitly version-specific and multi-model, unlike the narrower reports." }, "evaluator_used": true, "retried": false, @@ -2238,13 +2181,13 @@ "issue:44683|issue:46515", "issue:36010|issue:42915", "issue:43645|issue:43646", - "issue:41553|issue:44488", "issue:44910|issue:46858", + "issue:41553|issue:44488", "issue:43550|issue:44910", "issue:45823|issue:46515", "issue:42947|issue:47254", - "issue:45812|issue:46489", "issue:44661|issue:45823", + "issue:45812|issue:46489", "issue:44560|issue:44805", "issue:44877|issue:45030", "issue:38175|issue:43994", @@ -2253,233 +2196,245 @@ "issue:44661|issue:44805", "issue:42491|issue:43720", "issue:45672|issue:46899", - "issue:43632|issue:45834", "issue:41720|issue:42915", + "issue:43632|issue:45834", "issue:43317|issue:44756", "issue:42915|issue:44756", "issue:38175|issue:39692", "issue:43644|issue:44534", "issue:42491|issue:45276", - "issue:44960|issue:44977", "issue:44451|issue:44488", + "issue:44960|issue:44977", "issue:45507|issue:46468", "issue:45313|issue:45357", "issue:44863|issue:46846" ], - "prepared_review_unit_hash": "ec68ad0d3d0f48c9314d96e17d0887b87908ac8a9f8a380719f0095368258960", + "prepared_review_unit_hash": "59aed0fd7beafe2b3d1fda86286ad02d94daf15546e17e295a23ccdf0424726d", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 35965, - "estimated_input_tokens": 8992, - "estimated_eval_tokens": 18240 + "serialized_chars": 35964, + "estimated_input_tokens": 8991, + "estimated_eval_tokens": 18238 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous issue set with a few clear duplicate pairs. Strong matches are the Transformers v5 non-persistent-buffer regression, the Gemma4 4-bit multimodal dtype-cast bug, the SigLIP2 bad-output reports, and the custom-model initialization regression. Most other pairs are only loosely related at the subsystem level and should stay separate.", - "confidence": 0.77, - "canonical_issue_reason": "issue:43644 is the best canonical issue: it is a broad, clearly stated Transformers v5 regression and has an obvious duplicate in issue:44534 with the same exact failure mode.", + "summary": "Most pairs are not duplicates; they share a subsystem but describe different failures. I only accepted a small set of clear same-bug matches, mostly exact or near-exact duplicates.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43644 is the clearest representative of the non-persistent-buffer regression, and 44534 is effectively the same report with a reworded title.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the strongest representative issue in this set because it describes a concrete, versioned regression with a direct duplicate and a clean, reusable title.", + "best_issue_reason": "Issue 43644 is the strongest canonical issue here because it is precise, broadly phrased, and has an almost exact duplicate in 44534.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44683", "right": "issue:46515", "accept": false, - "reason": "Both mention torch-related incompatibility, but they are different version failures and do not describe the same bug." + "reason": "Different problems: torch>=2.9 flex_attention compilation vs a torch 2.4.1 incompatibility." }, { "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "Importing GenerationMixin and Qwen3-MoE LoRA incompatibility are unrelated failures." + "reason": "Unrelated failures: GenerationMixin import error vs Qwen3MoE FP8 config handling." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "Both report the same custom model initialization regression; the Jupyter notebook context in 43645 is just one manifestation." + "reason": "Same underlying custom-model initialization regression in Transformers v5; notebook mention is just a usage variant." }, { - "left": "issue:41553", - "right": "issue:44488", + "left": "issue:44910", + "right": "issue:46858", "accept": false, - "reason": "Different model-loading problems for different checkpoints; no evidence of the same root cause." + "reason": "Different code paths: flash-attention 3D position_ids crash vs static-cache seq-length arithmetic on non-CUDA backends." }, { - "left": "issue:44910", - "right": "issue:46858", + "left": "issue:41553", + "right": "issue:44488", "accept": false, - "reason": "One is a Flash Attention packed-sequence bug on Qwen3.5, the other is a static-cache/torch.compile backend issue." + "reason": "Both are model-loading complaints, but they concern different models and no shared concrete bug is evident." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Different models and different failure paths: SDPA/torch.compile vs Flash Attention/position_ids." + "reason": "Different failures in different attention stacks and models." }, { "left": "issue:45823", "right": "issue:46515", "accept": false, - "reason": "Device-map forward mismatch is unrelated to a torch version compatibility report." + "reason": "Device-map forward mismatch vs a general torch version incompatibility; not the same bug." }, { "left": "issue:42947", "right": "issue:47254", "accept": false, - "reason": "Related area, but one is ineffective gradient checkpointing while the other is an actual CheckpointError under ZeRO-3; not the same concrete bug." + "reason": "Both involve PEFT and checkpointing, but one is ineffective checkpointing while the other is a ZeRO-3 checkpoint error." }, { - "left": "issue:45812", - "right": "issue:46489", + "left": "issue:44661", + "right": "issue:45823", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model families and appear to come from different fixes." + "reason": "Tokenizer-mapping registration bug vs Gemma4 device placement issue." }, { - "left": "issue:44661", - "right": "issue:45823", + "left": "issue:45812", + "right": "issue:46489", "accept": false, - "reason": "Tokenizer-mapping helper failure and Gemma4 device mismatch are unrelated." + "reason": "Both are tokenizer regressions, but they affect different model families and appear to be separate bugs." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Different runtime errors in different code paths; no shared underlying bug is evident." + "reason": "Different symptom and path: video StopIteration vs tensor mask shape IndexError." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Both are config-loading failures, but the evidence is too generic to treat them as the same underlying issue." + "reason": "Both are config-validation load failures, but for different models and likely different root causes." }, { "left": "issue:38175", "right": "issue:43994", "accept": true, - "reason": "These are the same SigLIP2 bad-output problem on the same checkpoint; one reports zero probabilities, the other nonsensical outputs." + "reason": "Same SigLIP2 regression: both report incorrect/nonsensical outputs for google/siglip2-base-patch16-224." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "DeepSpeed ZeRO-3 index error is unrelated to the TOKENIZER_MAPPING_NAMES add-new-model-like failure." + "reason": "Bert/DeepSpeed ZeRO-3 init error is unrelated to add-new-model-like tokenizer mapping." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Different bug classes: ZeRO-3/BertModel indexing vs tokenizer-mapping helper behavior." + "reason": "DeepSpeed ZeRO-3 Bert init IndexError is unrelated to tokenizer mapping registration." }, { "left": "issue:44661", "right": "issue:44805", "accept": false, - "reason": "Unrelated failures: tokenizer-mapping helper vs mask-shape IndexError." + "reason": "Different bugs: tokenizer-mapping registration vs mask/tensor shape mismatch." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "LoRA hf4.x/hf5.x incompatibility and BitNet packed-weight loading are different code paths." + "reason": "Different model/version compatibility issues with different failure modes." }, { "left": "issue:45672", "right": "issue:46899", "accept": true, - "reason": "Both describe the same Gemma4 4-bit multimodal dtype-casting failure affecting audio/vision embedders." + "reason": "Same Gemma4 4-bit multimodal quantization bug; both point to unsafe dtype handling in embedders." }, { - "left": "issue:43632", - "right": "issue:45834", + "left": "issue:41720", + "right": "issue:42915", "accept": false, - "reason": "Transformers v5 _is_hf_initialized flag breakage and Kosmos2.5 OCR index error are unrelated." + "reason": "Different models and different runtime failures." }, { - "left": "issue:41720", - "right": "issue:42915", + "left": "issue:43632", + "right": "issue:45834", "accept": false, - "reason": "Auto device mapping cudaErrorAssert and Qwen3-MoE hf4.x/hf5.x incompatibility are different issues." + "reason": "Internal flag regression vs OCR input index error; no shared bug." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "Dequantized device_map offload failure and Strix Halo mmap OOM are not the same bug." + "reason": "Different resource/loading problems: dequantized gpu+cpu offload vs mmap OOM on Strix Halo." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "Different symptoms and different subsystems; only broadly related by large-model loading." + "reason": "Qwen3MoE FP8 failure is unrelated to mmap/OOM behavior." }, { "left": "issue:38175", "right": "issue:39692", "accept": false, - "reason": "Both involve SigLIP2, but 39692 is a compound docs-example issue with extra quantization problems, so it is not clearly the same bug as 38175." + "reason": "Same model family, but one is bad inference output and the other is a documentation example with model/processor and quantization issues." }, { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "Exact duplicate regression: both say Transformers v5 fills non-persistent buffers with junk." + "reason": "Near-exact duplicate: both report Transformers v5 filling non-persistent buffers with junk." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "LoRA checkpoint compatibility and Gemma4 resize_token_embeddings behavior are unrelated." + "reason": "LoRA/hf4.x vs hf5.x incompatibility is unrelated to Gemma4 embedding resize behavior." }, { - "left": "issue:44960", - "right": "issue:44977", + "left": "issue:44451", + "right": "issue:44488", "accept": false, - "reason": "Different models and failure modes; one is GLM5, the other is Qwen3.5 flash-attention generation." + "reason": "Both are loading failures, but for different checkpoints and without evidence of the same root cause." }, { - "left": "issue:44451", - "right": "issue:44488", + "left": "issue:44960", + "right": "issue:44977", "accept": false, - "reason": "Both are load failures, but the checkpoints and likely root causes are different and not shown to be the same bug." + "reason": "Different model families and distinct attention/runtime issues." }, { "left": "issue:45507", "right": "issue:46468", "accept": false, - "reason": "GraniteMoEHybrid invalid method call and Nemotron example breakage are unrelated." + "reason": "GraniteMoEHybrid invalid method call is unrelated to the Nemotron example failure." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Both are Qwen3.5 regressions, but one is ZeRO-3 weight loading and the other is incorrect visual-encoder keys on save_pretrained." + "reason": "Different Qwen3.5 bugs: ZeRO-3 weight loading vs incorrect save_pretrained visual encoder keys." }, { "left": "issue:44863", "right": "issue:46846", "accept": false, - "reason": "NemotronH checkpoint loading and Qwen3.5 TP-plan OOM/shape errors are different bugs." + "reason": "NemotronH checkpoint loading and Qwen3.5 TP-plan issues are unrelated." + }, + { + "left": "issue:44910", + "right": "issue:44977", + "accept": false, + "reason": "Different models and different flash-attention/runtime symptoms." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 BertModel IndexError; the pretrained/non-pretrained detail does not change the underlying bug." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded overall. The canonical issue choice is well supported by the titles, and the accepted soft pairs are limited to clearly matching regressions or same-checkpoint output failures. The rejected pairs are generally conservative and do not overclaim duplicates from loose subsystem similarity." + "feedback": "The summary is grounded in the packet, and the accepted pairs are all reasonably conservative near-duplicates. I don\u2019t see an overstatement strong enough to reject this review." }, "evaluator_used": true, "retried": false, @@ -2568,8 +2523,8 @@ "issue:43425|issue:43957", "issue:44291|issue:44756", "issue:45362|issue:45406", - "issue:41720|issue:44155", "issue:45663|issue:45759", + "issue:41720|issue:44155", "issue:39692|issue:44960", "issue:44910|issue:45588", "issue:44402|issue:44488", @@ -2579,223 +2534,223 @@ "issue:42947|issue:45704", "issue:45663|issue:45859", "issue:44155|issue:45823", - "issue:44079|issue:46829", - "issue:43957|issue:44756" + "issue:43957|issue:44756", + "issue:44079|issue:46829" ], - "prepared_review_unit_hash": "a56a88c77d8a5241126c9fce97686b2381347f95372aaab73dd92bf2bba04434", + "prepared_review_unit_hash": "15b8449bdb0ffdf163e8c12d5162579de274389354afcda697057e7ea401bc69", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37283, - "estimated_input_tokens": 9321, - "estimated_eval_tokens": 18898 + "serialized_chars": 37280, + "estimated_input_tokens": 9320, + "estimated_eval_tokens": 18896 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: it contains several unrelated model-loading, tokenizer, quantization, and training bugs. A few pairs are superficially similar, but none are safe duplicate merges from the available evidence. The strongest near-match is around resize_token_embeddings, but it is still model-specific rather than the same concrete bug.", - "confidence": 0.44, - "canonical_issue_reason": "issue:46459 is the best broad representative of the loading/dtype regression thread: it describes a concrete, cross-model failure mode in composite multimodal checkpoint loading and has the clearest generalized scope among the set.", + "summary": "The set is largely heterogeneous; none of the soft pairs are safe duplicate merges based on the visible titles alone. Most pairs share only a broad subsystem (loading, quantization, tokenization, training) but differ in model, code path, or failure mode.", + "confidence": 0.91, + "canonical_issue_reason": "If forced to pick one representative issue, issue:43404 is the clearest, most code-path-specific bug report in the set: it names the affected model, the concrete symptom (untied lm_head), and points to a likely fix location in weight tying logic.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46459 is the most generally useful issue to anchor this mixed cluster because it is high-signal, broadly applicable, and tied to an explicit Transformers loading regression rather than a single narrow model quirk.", + "best_issue_reason": "issue:43404 is the best standalone candidate because it is narrowly scoped, reproducible, and likely to produce a focused fix; the rest are either broader environment regressions or model-specific variations that are not duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46459", "right": "issue:46512", "accept": false, - "reason": "Related dtype/loading symptoms, but one is about composite checkpoint loading while the other is about from_config leaving config.dtype stale; different code paths." + "reason": "Both mention dtype-related model loading, but one is about ignored dtype during checkpoint loading and the other about stale config.dtype after from_config; different bugs and code paths." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Both involve SigLIP2, but one is a docs/example mismatch and quantization failure while the other is a model/pipeline output regression." + "reason": "Both involve SigLIP2, but one is a doc example with processor/model mismatch and quantization failure while the other is incorrect outputs from AutoModel/pipeline; not the same underlying defect." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer-loading failures, but for different models and likely different tokenizer files/code paths." + "reason": "Both are tokenizer-loading regressions, but they affect different models and likely different tokenizer files/code paths." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Same release window, but one is junk non-persistent buffers and the other is notebook custom-model initialization; not the same bug." + "reason": "Both are Transformers 5.0 complaints, but one is about non-persistent buffers and the other about notebook-defined custom models; unrelated failures." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Different models and failures: Qwen3 MoE weight conversion vs GLM-4.7 load correctness in tests." + "reason": "Both are model-loading issues, but they concern different models and different failure modes (Qwen3 MoE conversion vs GLM-4.7-Flash unit-test loading)." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Both mention memory/quantization, but one is training memory usage and the other is offloading behavior; mechanisms differ." + "reason": "One is Qwen3 MoE training memory usage, the other is quantization/offloading behavior; broad memory theme only, not the same bug." }, { "left": "issue:42491", "right": "issue:45335", "accept": false, - "reason": "LoRA/Qwen3 MoE version incompatibility is unrelated to a resize_token_embeddings propagation bug in t5gemma." + "reason": "Different models and different symptoms; both are embedding-related but not the same concrete bug." }, { "left": "issue:42491", "right": "issue:45292", "accept": false, - "reason": "Different underlying problems: LoRA model compatibility vs resize_token_embeddings not updating output embeddings." + "reason": "Same broad resize/token-embedding theme, but the reports target different model families and embedding surfaces." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Separate model-loading failures for different model families; no concrete shared bug." + "reason": "Different models with different loading failures; no shared concrete code path is evident." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "Both are load/save related, but one is extra_state handling and the other is unexpected _is_hf_initialized in init_empty_weights." + "reason": "Both concern loading/save mechanics, but one is extra_state handling and the other is init_empty_weights/_is_hf_initialized argument handling; separate issues." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both mention memory pressure under quantization, but one is training inefficiency and the other is a reserved-memory/OOM regression." + "reason": "Both mention memory pressure, but one is training-memory inefficiency and the other is int4 quantization reserved-memory/OOM regression." }, { "left": "issue:44960", "right": "issue:46050", "accept": false, - "reason": "Completely different models and failure modes; no evidence of the same bug." + "reason": "Different model families and different persistence problems; no evidence of the same bug." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but one overrides pre-tokenization behavior while the other returns wrong token IDs for Granite." + "reason": "Both are tokenizer regressions in v5, but one changes pre-tokenizer behavior for DeepSeek-family models and the other yields wrong IDs for Granite; distinct failures." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Trainer checkpoint math and optimizer argument usage are unrelated issues." + "reason": "Trainer resume/max_steps logic is unrelated to optimizer argument handling." }, { "left": "issue:45276", "right": "issue:45335", "accept": false, - "reason": "Very similar symptom, but they affect different model implementations and embedding paths; not enough evidence for one underlying bug." + "reason": "Both are resize_token_embeddings complaints, but they affect different architectures and different embedding endpoints; too little evidence for one underlying bug." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch version compatibility vs meta-device model loading are different problems." + "reason": "Torch version incompatibility and meta-device loading failures are different regressions." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "Different failures: init_empty_weights argument handling vs mmap-induced OOM on Strix Halo." + "reason": "init_empty_weights TypeError is unrelated to Strix Halo mmap OOM behavior." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Different models and different crashes: Qwen3.5 chat vs Gemma4Processor missing _tokenizer." - }, - { - "left": "issue:41720", - "right": "issue:44155", - "accept": false, - "reason": "CUDA assert on auto device mapping is unrelated to AudioFlamingo3 batched inference token leakage." + "reason": "Different models and different crash causes; no duplicate relationship is apparent." }, { "left": "issue:45663", "right": "issue:45759", "accept": false, - "reason": "FSDP2 shared_kv_states rebuilding and AutoModelForCausalLM text_config unwrapping are distinct code-path bugs." + "reason": "FSDP2 shared_kv_states bug and composite-config unwrapping bug are separate code paths." }, { - "left": "issue:39692", + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different models and failure types; one is CUDA assert with device mapping, the other is batched inference leakage." + }, + { + "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "No shared bug: SigLIP2 docs/example issues vs a separate GLM5 item with no matching failure description." + "reason": "No substantive overlap beyond both being model-specific issues." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Different crash mechanisms: Flash Attention illegal access from position_ids vs AttributeError on s_aux=None." + "reason": "Both involve flash attention, but one is a Qwen3.5 packed-sequence/position_ids crash and the other is an AttributeError on sink-less models; not the same bug." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "One is vocab-size mismatch for pad/unk tokens, the other is a tokenizer loading failure for a different model." + "reason": "Different tokenizer/model-loading issues with no shared concrete root cause." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Both involve untied lm_head weights, but they are different architectures with different forward paths and symptoms." + "reason": "Both are lm_head tying problems, but they affect different model families and likely different initialization paths." }, { "left": "issue:46897", "right": "issue:46901", "accept": false, - "reason": "The titles suggest a similar double-shift pattern, but they are model-specific training-loss bugs with insufficient evidence of the same underlying change." + "reason": "Both describe a training-loss double-shift pattern, but they are for different models and no shared implementation bug is confirmed." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Both are memory-related quantization issues, but one is offloading behavior and the other is increased reserved CUDA memory under int4." + "reason": "Both touch quantization/memory, but one is offloading behavior and the other is reserved-memory growth under int4 quantization; different regressions." }, { "left": "issue:42947", "right": "issue:45704", "accept": false, - "reason": "Gradient checkpointing with PEFT LoRA and an apex RMSNorm memory leak are unrelated." + "reason": "Gradient checkpointing/PEFT behavior is unrelated to a T5 RMSNorm memory leak." }, { "left": "issue:45663", "right": "issue:45859", "accept": false, - "reason": "Different subsystems: FSDP2 attention state rebuild vs CPU-offload compatibility in rotary embedding." + "reason": "Different model families and different runtime issues (FSDP2 attention state vs CPU-offload incompatibility)." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "Audio track embedding/token leakage and Gemma4 PLE device mismatch are unrelated bugs." + "reason": "Both are multimodal/runtime issues, but one is track-level leakage in AudioFlamingo3 and the other is device mismatch in Gemma4 PLE; not duplicates." }, { - "left": "issue:44079", - "right": "issue:46829", + "left": "issue:43957", + "right": "issue:44756", "accept": false, - "reason": "Different output object issues in different models; one is generic ModelOutput key assignment and the other is a VideoPrism classifier output field mix-up." + "reason": "Meta-device loading and mmap/OOM are separate platform regressions." }, { - "left": "issue:43957", - "right": "issue:44756", + "left": "issue:44079", + "right": "issue:46829", "accept": false, - "reason": "Meta-device model loading and mmap OOM on Strix Halo are separate failure modes." + "reason": "ModelOutput key assignment and ImageClassifierOutput field mispopulation are different bugs." } ] }, @@ -2869,19 +2824,19 @@ "issue:47332" ], "soft_pairs": [ - "issue:43335|issue:45663", "issue:44863|issue:46468", + "issue:43335|issue:45663", "issue:44479|issue:44877", "issue:44387|issue:44756", - "issue:36010|issue:39404", "issue:45276|issue:45292", + "issue:36010|issue:39404", "issue:44292|issue:44912", "issue:44164|issue:44756", "issue:43957|issue:47332", "issue:43381|issue:44928", "issue:45588|issue:45663", - "issue:43645|issue:44164", "issue:43425|issue:47228", + "issue:43645|issue:44164", "issue:42796|issue:46326", "issue:44589|issue:44756", "issue:43425|issue:43828", @@ -2899,14 +2854,14 @@ "issue:46315|issue:46491", "issue:39404|issue:45701" ], - "prepared_review_unit_hash": "1d5a581ef1a6984d235c3797c8a79bd38df3d680dbd3c22144af5fdc36387933", + "prepared_review_unit_hash": "d856ebdf2ed388ab3799784b0e90df72012e2a87ba02dbf6b21291105a07c8e6", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, @@ -2921,192 +2876,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of unrelated Transformers issues. The only clear duplicate family is the resize_token_embeddings regression across model-specific embeddings; most other soft pairs should stay separate.", - "confidence": 0.79, - "canonical_issue_reason": "Issue #45292 is the best representative of the only clear duplicate family: it states the core resize_token_embeddings/output_embeddings bug in a generic way, while the nearby model-specific reports are narrower manifestations.", + "summary": "This cluster is mostly heterogeneous and does not form one clear duplicate set. I only see two plausible duplicate mini-clusters: the resize_token_embeddings regression family, and the training-loss double-shift family. Everything else looks like a separate bug or regression.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue for the entire set; the items split into unrelated bugs. If narrowing to the only clear duplicate family, issue 45292 is the broadest representative for the resize_token_embeddings regressions.", "canonical_pr_reason": null, - "best_issue_reason": "#45292 is the most canonical issue for the shared embedding-resize regression because it is the broadest, least model-specific description of the bug.", + "best_issue_reason": "Issue 45292 is the cleanest issue to anchor the resize_token_embeddings subset because it is the most generic description and matches the model-specific variants.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43335", - "right": "issue:45663", + "left": "issue:44863", + "right": "issue:46468", "accept": false, - "reason": "Different bugs: sparse-layer construction in SwitchTransformers vs a Gemma4 FSDP2 shared_kv_states KeyError." + "reason": "Both are Nemotron-related, but one is checkpoint loading and the other is an example breakage; not the same concrete bug." }, { - "left": "issue:44863", - "right": "issue:46468", + "left": "issue:43335", + "right": "issue:45663", "accept": false, - "reason": "Both are Nemotron-related, but one is a checkpoint-loading failure and the other is a broken example; no clear same code-path bug." + "reason": "Different models and different failure modes; no shared code-path evidence." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Unrelated regressions: Qwen VL video-input handling vs Granite speech config strictness." + "reason": "Both are model-loading regressions, but one is Qwen-VL video input and the other is Granite Speech config strictness." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Both mention OOM, but one is int4 CUDA reserved memory regression and the other is mmap-on-Strix-Halo behavior; different root causes." - }, - { - "left": "issue:36010", - "right": "issue:39404", - "accept": false, - "reason": "ImportError for GenerationMixin and Whisper return_language pipeline failure are unrelated." + "reason": "Both mention OOM, but the causes differ: int4 quantization memory growth vs Strix Halo mmap behavior." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Same underlying resize_token_embeddings regression: model-specific missing propagation to output/embedding layers." + "reason": "Same underlying resize_token_embeddings propagation bug; 45276 looks like a Gemma4-specific manifestation of the broader 45292 issue." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "ImportError for GenerationMixin and Whisper pipeline return_language are unrelated." }, { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Different quantization/load problems: Qwen NVFP4 runtime error vs MXFP4 falling back to bf16." + "reason": "Different quantization/model-loading regressions with different symptoms and code paths." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "save/from_pretrained extra_state handling is unrelated to mmap/OOM behavior." + "reason": "save/from_pretrained extra_state handling is unrelated to Strix Halo mmap OOM." }, { "left": "issue:43957", "right": "issue:47332", "accept": false, - "reason": "Meta-device loading breakage and disk-offloaded saving failure are distinct serialization/offloading issues." + "reason": "Meta-device model loading and disk-offloaded saving are different lifecycle paths." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Gradient checkpointing in eval mode is unrelated to Qwen3.5 NaN/BF16 SDPA fallback." + "reason": "Gradient checkpointing eval-mode restriction is unrelated to the RLHF NaN/SDPA fallback bug." }, { "left": "issue:45588", "right": "issue:45663", "accept": false, - "reason": "Flash-attention s_aux=None crash and Gemma4 FSDP2 KeyError are different attention/training failures." + "reason": "Flash-attention sink handling and Gemma4 FSDP2 shared_kv_states are separate issues." }, { - "left": "issue:43645", - "right": "issue:44164", + "left": "issue:43425", + "right": "issue:47228", "accept": false, - "reason": "Custom model initialization in Jupyter is not the same bug as extra_state save/load handling." + "reason": "Torch version incompatibility and Sam3 bf16 compilation regression are not the same bug." }, { - "left": "issue:43425", - "right": "issue:47228", + "left": "issue:43645", + "right": "issue:44164", "accept": false, - "reason": "Generic Torch 2.10 incompatibility is too broad and not the same concrete regression as Sam3 bf16 compilation." + "reason": "Custom-model notebook initialization and extra_state save/load failures are distinct." }, { "left": "issue:42796", "right": "issue:46326", "accept": false, - "reason": "Tokenizer BPE-marker handling and mistral_common continue_final_message are different tokenizer behaviors." + "reason": "Tokenizer BPE-marker handling and mistral_common continue_final_message behavior are different issues." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Float8 storage deserialization failure is unrelated to mmap/OOM on Strix Halo." + "reason": "Float8 storage lookup failure is unrelated to Strix Halo mmap OOM." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch version incompatibility is too broad; Phi-tiny-MoE autocast dtype mismatch is a specific model/runtime bug." + "reason": "Torch 2.10 compatibility and Phi-tiny-MoE autocast dtype mismatch are unrelated." }, { "left": "issue:45292", "right": "issue:45335", "accept": true, - "reason": "Same resize_token_embeddings bug expressed on another model-specific embedding/output path." + "reason": "Both describe the same resize_token_embeddings bug affecting downstream embedding modules in different model variants." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "GLM-4.7-Flash test loading issue is unrelated to Qwen3.5 save_pretrained checkpoint corruption." + "reason": "GLM-4.7-Flash loading failure and Qwen3.5 save_pretrained regression are separate model-specific problems." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom model initialization in notebooks is unrelated to Trainer max_steps recomputation on checkpoint resume." + "reason": "Custom model initialization breakage and Trainer max_steps miscalculation are unrelated." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "Tokenizer vocab-size mismatch and ScandiBERT load failure are not clearly the same underlying bug." + "reason": "Both involve model/tokenizer loading, but the titles point to different concrete failures and no shared bug is clear." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "Read-only permission propagation from save_pretrained is a different issue from loading the wrong custom module." + "reason": "Read-only permission propagation and wrong custom-module loading after save_pretrained are different defects." }, { "left": "issue:43425", "right": "issue:46869", "accept": false, - "reason": "Broad Torch incompatibility is not the same concrete bug as YOLO multi-GPU device_map auto failure." + "reason": "Torch 2.10 incompatibility is unrelated to YOLO device_map auto multi-GPU failure." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Torch compatibility complaints do not match the specific Mistral-3 image-preprocessor dtype mismatch." + "reason": "Torch version compatibility and Mistral-3 image preprocessor dtype mismatch are different problems." }, { "left": "issue:44683", "right": "issue:45823", "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 is unrelated to Gemma4 device-map placement issues." + "reason": "Compiled flex_attention regression and Gemma4 PLE device mismatch are unrelated." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "PagedAttentionCache linear_attention crash and non-template-node compile error are different code paths." + "reason": "PagedAttentionCache linear_attention crash and non-template node compilation error are not the same code-path." }, { "left": "issue:46894", "right": "issue:46901", - "accept": false, - "reason": "Both are training-loss double-shift reports, but they reference different models and no shared concrete implementation is shown." + "accept": true, + "reason": "Both are the same training-loss double-shift pattern: labels are shifted twice instead of using the unshifted labels." }, { "left": "issue:45812", "right": "issue:46315", "accept": false, - "reason": "Granite token ID regression and slow convert_tokens_to_ids performance are different tokenizer issues." + "reason": "Wrong Granite token IDs and slow convert_tokens_to_ids are separate tokenizer regressions." }, { "left": "issue:46315", "right": "issue:46491", "accept": false, - "reason": "Performance regression in token ID conversion is unrelated to CodeLlama leading-space round-trip behavior." + "reason": "Performance regression in token lookup is unrelated to CodeLlama spacing round-trip behavior." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper pipeline return_language failure and version-dependent tokenization changes are not the same bug." + "reason": "Whisper return_language pipeline breakage and tokenizer behavior changes are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded and conservative overall. The canonical issue choice for the resize_token_embeddings family is supported by the packet, and the accepted pairs (45276\u219445292 and 45292\u219445335) look like the same bug class with model-specific manifestations. The other false verdicts are reasonably cautious and do not overstate shared root causes." + "feedback": "The summary is grounded in the packet: the cluster is mostly heterogeneous, and the only defensible duplicate families are the resize_token_embeddings trio (45276/45292/45335) and the training-loss double-shift pair (46894/46901). The soft-edge rejections are conservative, and the two acceptances are supported by closely aligned issue titles." }, "evaluator_used": true, "retried": false, @@ -3134,6 +3089,7 @@ "issue:42915", "issue:43404", "issue:43452", + "issue:43541", "issue:43596", "issue:43611", "issue:43632", @@ -3181,10 +3137,10 @@ "issue:44589|issue:45084", "issue:44164|issue:45759", "issue:43994|issue:44079", - "issue:44265|issue:44898", "issue:40990|issue:47331", - "issue:39404|issue:46665", + "issue:44265|issue:44898", "issue:36032|issue:43452", + "issue:39404|issue:46665", "issue:43901|issue:44509", "issue:44991|issue:45081", "issue:46050|issue:46326", @@ -3197,218 +3153,231 @@ "issue:43596|issue:44683", "issue:44756|issue:47333", "issue:43856|issue:44756", - "issue:43632|issue:47050", "issue:44488|issue:44987", "issue:36032|issue:42915", + "issue:43632|issue:47050", "issue:44488|issue:45081", "issue:43716|issue:45084", "issue:43746|issue:47331", "issue:42796|issue:43404", "issue:43611|issue:43644", - "issue:44661|issue:44683" + "issue:44661|issue:44683", + "issue:43541|issue:45084" ], - "prepared_review_unit_hash": "a51b214fb2e75cfeb48be4a821e604541b1c408bf34725ce050dc1ed5f5be908", + "prepared_review_unit_hash": "412b41472214e37c520b0833cb23bc1f6f475b68a297693892a09f8091cd68f2", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 30, - "serialized_chars": 36376, - "estimated_input_tokens": 9094, - "estimated_eval_tokens": 18444 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37279, + "estimated_input_tokens": 9320, + "estimated_eval_tokens": 18896 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items are a heterogeneous set of mostly unrelated Transformers issues. None of the soft pairs look like the same underlying bug/change closely enough to merge.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue fits the cluster: the issues span distinct regressions across tokenizers, model loading, quantization, multimodal models, and docs.", + "summary": "This cluster is mostly heterogeneous: many issues are about distinct v5 regressions, model-specific loading bugs, or unrelated runtime errors. I did not find any soft pair that looks like the same underlying bug/change; the closest common theme is general Transformers v5 model-loading/initialization breakage.", + "confidence": 0.78, + "canonical_issue_reason": "issue:43611 is the broadest and most central representative of the only semi-coherent subtheme here: v5 model-loading/init regressions. Even so, the cluster is not a true duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue; the set is too diverse and the soft pairs do not collapse into one representative defect.", + "best_issue_reason": "issue:43611 is the best single representative because it is general, well-scoped, and closest to the shared v5 loading/initialization regression theme seen in a few other issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Both are v5-era model-init/load regressions, but one is about base_model_prefix loading and the other about notebook-defined custom models; different failure modes and fixes." + "reason": "Both are v5 model-loading breakages, but one is base_model_prefix handling and the other is custom model initialization in notebooks; different failure paths." }, { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Different subsystems and symptoms: video input regression for Qwen VL models vs a generic mask/tensor IndexError." + "reason": "Video-input regression in Qwen VL models vs a mask/tensor shape IndexError; unrelated code paths and symptoms." }, { "left": "issue:44589", "right": "issue:45084", "accept": false, - "reason": "Unrelated failures: missing Float8 storage object vs compiler error about non-template nodes." + "reason": "Missing Float8 storage support vs a compiler/template-node TypeError; different subsystems and root causes." }, { "left": "issue:44164", "right": "issue:45759", "accept": false, - "reason": "One is save/from_pretrained handling of extra_state; the other is from_config unwrapping text_config for composite Qwen configs." + "reason": "extra_state save/load handling vs unwrapping text_config in composite multimodal configs; different APIs and bugs." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "SigLIP inference corruption and ModelOutput key assignment are different code paths with no clear shared bug." - }, - { - "left": "issue:44265", - "right": "issue:44898", - "accept": false, - "reason": "torch.export failures for torch_compilable_check are not the same as Perceiver image-resolution handling." + "reason": "Model quality/regression report vs ModelOutput key assignment logic; not the same underlying issue." }, { "left": "issue:40990", "right": "issue:47331", "accept": false, - "reason": "High perplexity on a language model is not the same as a MetalConfig quantization batched-generate corruption bug." + "reason": "Perplexity on WikiText-2 vs MetalConfig batched quantized generation corruption; unrelated problems." }, { - "left": "issue:39404", - "right": "issue:46665", + "left": "issue:44265", + "right": "issue:44898", "accept": false, - "reason": "Whisper pipeline return_language regression and from_pretrained-in-classes inconsistency are distinct issues." + "reason": "torch.export failure around torch_compilable_check vs Perceiver non-default resolution interpolation; different model/export paths." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "Tokenizer initialization conflict in T5 is unrelated to gguf_file breaking AutoTokenizer/AutoModel loading." + "reason": "Tokenizer method-name conflict vs gguf_file loading breakage; distinct tokenizer/model loading issues." + }, + { + "left": "issue:39404", + "right": "issue:46665", + "accept": false, + "reason": "Whisper pipeline return_language regression vs from_pretrained behavior inside Python classes; unrelated bugs." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs-related, but one concerns return_all_scores behavior and the other removed pipeline tasks; different documentation changes." + "reason": "Both are docs-related, but one is about return_all_scores and the other about removed pipeline tasks; different documentation mismatches." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Both involve tokenizer loading, but they affect different model families and different root causes (est-roberta load failure vs Mistral regex patch crash)." + "reason": "Different tokenizer failures on different models: est-roberta loading vs Mistral regex patch crash." }, { "left": "issue:46050", "right": "issue:46326", "accept": false, - "reason": "Quantized save/load breakage via sinq is unrelated to continue_final_message behavior for mistral_common tokenizers." + "reason": "Quantized sinq save/load breakage vs continue_final_message on mistral_common tokenizers; different features and code paths." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "CamemBERT masked-LM prediction regression and est-roberta tokenizer loading failure are different bugs." + "reason": "CamemBERT masked-LM prediction regression vs tokenizer loading failure for a different model; not the same bug." }, { "left": "issue:43906", "right": "issue:46326", "accept": false, - "reason": "An isolated reproduction of another issue is not the same as the mistral_common continue_final_message bug." + "reason": "One is a reproduction of an older issue, the other is a Mistral-common tokenizer behavior bug; not duplicates." }, { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "Generic reproduction report vs Qwen3.5 PagedAttentionCache linear_attention crash; no shared concrete defect." + "reason": "Isolated reproduction of 38071 vs PagedAttentionCache linear_attention crash; different failures." }, { "left": "issue:39404", "right": "issue:44912", "accept": false, - "reason": "Whisper pipeline language return regression and MXFP4 quantization fallback on git-oss-20b are unrelated." + "reason": "Whisper pipeline return_language regression vs MXFP4 quantization load fallback; unrelated functionality." }, { "left": "issue:45823", "right": "issue:46899", "accept": false, - "reason": "Same model family, but device_map auto device mismatch and unsafe dtype cast in multimodal embedders are separate bugs." + "reason": "Gemma4 device_map mismatch in forward vs unsafe dtype cast in multimodal embedders; different bugs." }, { "left": "issue:44164", "right": "issue:46821", "accept": false, - "reason": "extra_state save/load failure and TimesFM window_size AttributeError are unrelated APIs and code paths." + "reason": "extra_state persistence bug vs TimesFM window_size AttributeError; unrelated APIs." }, { "left": "issue:43596", "right": "issue:44683", "accept": false, - "reason": "deepspeed ZeRO-3/BertModel index error is not the same as torch>=2.9 flex_attention compilation failure." + "reason": "DeepSpeed ZeRO3/BertModel init IndexError vs flex_attention compile failure on torch>=2.9; different stacks." }, { "left": "issue:44756", "right": "issue:47333", "accept": false, - "reason": "Strix Halo mmap OOM mitigation and disk-offloaded Qwen save failure are different storage/memory problems." + "reason": "Strix Halo mmap/OOM workaround vs saving a disk-offloaded Qwen3 model; different memory/save behaviors." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "Qwen3 MoE training memory inefficiency is not the same as disabling mmap to avoid OOM on Strix Halo." - }, - { - "left": "issue:43632", - "right": "issue:47050", - "accept": false, - "reason": "_is_hf_initialized flag breakage and ONNX attention dtype mismatch are unrelated issues." + "reason": "Qwen3 MoE training memory inefficiency vs Strix Halo mmap/OOM; same broad area, but not the same bug." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "Both are loading regressions, but they target different models and likely different root causes." + "reason": "Two loading failures, but for different models and different root causes; not one concrete bug." }, { - "left": "issue:44488", - "right": "issue:45081", + "left": "issue:43611", + "right": "issue:43644", "accept": false, - "reason": "cjvt/sleng-bert loading failure is unrelated to the Mistral regex patch crash." + "reason": "Both are v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization; separate bugs." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Mistral-3 image preprocessor/model dtype mismatch and template-node compile failure do not share a concrete code path." + "reason": "Mistral-3 image preprocessing dtype mismatch vs a compile-time template-node error; unrelated." }, { "left": "issue:43746", "right": "issue:47331", "accept": false, - "reason": "GraniteSpeech PEFT checkpoint loading and MetalConfig batched generate corruption are different failure domains." + "reason": "PEFT adapter checkpoint loading vs disk-offloaded quantized generation save/load; different features and symptoms." }, { "left": "issue:42796", "right": "issue:43404", "accept": false, - "reason": "Tokenizer BPE-marker handling in Ministral-3 and lm_head weight tying in Mistral3ForConditionalGeneration are separate bugs." + "reason": "Tokenizer BPE marker handling vs lm_head weight tying in Mistral3 conditional generation; different code paths." }, { - "left": "issue:43611", - "right": "issue:43644", + "left": "issue:44661", + "right": "issue:44683", "accept": false, - "reason": "Both are v5 regressions, but base_model_prefix loading and non-persistent buffer initialization are different underlying defects." + "reason": "Model-registration/tokenizer-mapping bug vs flex_attention compile failure; unrelated." }, { - "left": "issue:44661", - "right": "issue:44683", + "left": "issue:43541", + "right": "issue:45084", "accept": false, - "reason": "Model registration logic in add-new-model-like is unrelated to flex_attention compilation on torch>=2.9." + "reason": "Mixtral grouped_mm tracing error vs non-template-node compilation error; distinct compiler/runtime failures." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer load conflict vs Qwen3Moe FineGrainedFP8Config failure; unrelated." + }, + { + "left": "issue:43632", + "right": "issue:47050", + "accept": false, + "reason": "_is_hf_initialized flag regression vs ONNX export dtype mismatch in Canine attention; different problems." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Model loading failure for sleng-bert vs Mistral regex patch crash; different models and mechanisms." } ] }, @@ -3445,12 +3414,12 @@ "issue:43421", "issue:43452", "issue:43482", - "issue:43541", "issue:43582", "issue:43632", "issue:43644", "issue:43645", "issue:43646", + "issue:43792", "issue:43873", "issue:43931", "issue:44206", @@ -3474,6 +3443,7 @@ "issue:46326", "issue:46402", "issue:46489", + "issue:46869", "issue:46894", "issue:46897", "issue:47013", @@ -3481,238 +3451,238 @@ "issue:47331" ], "soft_pairs": [ - "issue:44462|issue:46489", - "issue:43541|issue:45084", "issue:45479|issue:46894", - "issue:43299|issue:43931", + "issue:44462|issue:46489", "issue:46285|issue:46402", + "issue:43299|issue:43931", "issue:42915|issue:44530", - "issue:43873|issue:47331", - "issue:45479|issue:46897", "issue:44451|issue:44987", + "issue:45479|issue:46897", + "issue:43873|issue:47331", "issue:36032|issue:43645", "issue:44530|issue:46304", - "issue:40990|issue:43421", "issue:44737|issue:45859", + "issue:40990|issue:43421", "issue:40444|issue:46326", "issue:34567|issue:40444", "issue:42175|issue:46178", "issue:43644|issue:43646", "issue:45084|issue:45507", "issue:43582|issue:43632", - "issue:39404|issue:44671", "issue:44206|issue:44220", "issue:44987|issue:44991", + "issue:39404|issue:44671", "issue:43452|issue:43482", "issue:43482|issue:47312", "issue:44811|issue:45584", - "issue:43208|issue:47013", - "issue:43582|issue:47331", "issue:43262|issue:47331", - "issue:42915|issue:45084" + "issue:43582|issue:47331", + "issue:43208|issue:47013", + "issue:42915|issue:45084", + "issue:43792|issue:46869" ], - "prepared_review_unit_hash": "6c8d63364ba23af06637813723b2e373971bbeac00dc9efdc05217d451c26694", + "prepared_review_unit_hash": "7c64ace01ee21771d6b863b48928221295bbbee503bc8e961a8b95c2eba779f5", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 35835, - "estimated_input_tokens": 8959, - "estimated_eval_tokens": 18174 + "serialized_chars": 36314, + "estimated_input_tokens": 9079, + "estimated_eval_tokens": 18414 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a heterogeneous collection of unrelated Transformers regressions and model-specific bugs. None of the soft pairs look like the same underlying bug/change closely enough to merge.", - "confidence": 0.78, - "canonical_issue_reason": "issue:43646 is the broadest, most umbrella-like report in the set ('breaks custom model initialization') and is the best issue anchor among these heterogeneous regressions.", + "summary": "This cluster is mostly heterogeneous Transformers regressions with only loose topical overlap. None of the soft pairs look like safe duplicates under a strict triage standard, so I would not merge any of them. The closest subgroup is the v5/custom-model-initialization cluster, where issue 43646 is the best representative anchor.", + "confidence": 0.31, + "canonical_issue_reason": "issue:43646 is the broadest, most generic report in the only clearly related subgroup: a Transformers v5 regression breaking custom model initialization. It is the best umbrella-style issue among the cluster, even though the overall cluster is noisy and not a tight duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43646 is the most generally scoped issue and could plausibly subsume nearby custom-init regressions better than the more model-specific reports.", + "best_issue_reason": "issue:43646 is the most suitable canonical issue because it states the core user-facing failure in the most general terms and is likely to subsume narrower initialization-related reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44462", - "right": "issue:46489", + "left": "issue:45479", + "right": "issue:46894", "accept": false, - "reason": "Both are tokenizer-loading regressions, but one is about AutoTokenizer ignoring tokenizer.json and the other is a DeepSeek-Coder tokenizer output bug tied to a prior fix gap; different code paths and failure modes." + "reason": "Both mention training-loss anomalies, but one is a zero-loss classification bug and the other is a sequence-shift bug in ASR training; different model classes and code paths." }, { - "left": "issue:43541", - "right": "issue:45084", + "left": "issue:44462", + "right": "issue:46489", "accept": false, - "reason": "Both mention compiler/tracing errors, but the concrete failures are different: grouped_mm tracing in Mixtral vs a template-node compilation TypeError." + "reason": "Both involve tokenizer behavior, but one is repository tokenizer loading and the other is a specific DeepSeek-Coder output regression after a prior fix; not the same bug." }, { - "left": "issue:45479", - "right": "issue:46894", + "left": "issue:46285", + "right": "issue:46402", "accept": false, - "reason": "One is a zero-loss classification bug with num_labels=1; the other is a training-loss double-shift issue in CohereASR. Similar theme, different bug." + "reason": "Different failures: CLIP text model structure inconsistency vs Sam3 text encoder weights not loading. No shared concrete code-path bug." }, { "left": "issue:43299", "right": "issue:43931", "accept": false, - "reason": "One is an HF initialization flag regression; the other is a Qwen3-VL weight-shape mismatch during loading. Not the same issue." - }, - { - "left": "issue:46285", - "right": "issue:46402", - "accept": false, - "reason": "These are unrelated model-loading regressions affecting different architectures and failure points." + "reason": "Both are Qwen-related loading issues, but one is a MoE loader breakage and the other is a weight-shape mismatch on a different VL model; too different to merge." }, { "left": "issue:42915", "right": "issue:44530", "accept": false, - "reason": "FP8 quantization failure in Qwen3Moe vs PagedAttentionCache linear_attention group-type crash in Qwen3.5 models; different subsystems and symptoms." + "reason": "Qwen3Moe FP8 loading failure and Qwen3.5 paged-attention cache crash are distinct bugs in different subsystems." }, { - "left": "issue:43873", - "right": "issue:47331", + "left": "issue:44451", + "right": "issue:44987", "accept": false, - "reason": "Offloading with quantization and MetalConfig batched-generate corruption are different quantization/offload bugs, not one concrete code-path problem." + "reason": "Both are loading failures, but one is a model-specific load problem and the other is a transformers>=5.1 incompatibility for a different model; not the same underlying defect." }, { "left": "issue:45479", "right": "issue:46897", "accept": false, - "reason": "Both involve loss computation, but one is a sequence-classification zero-loss bug and the other is Florence2 training-loss double-shift; not the same defect." + "reason": "Same general training-loss theme, but these are different model families and different shifting behavior; not one concrete bug." }, { - "left": "issue:44451", - "right": "issue:44987", + "left": "issue:43873", + "right": "issue:47331", "accept": false, - "reason": "Different model-specific loading failures; no evidence they share the same underlying tokenizer/model code path." + "reason": "Offloading with quantization is a broad runtime issue, while the other is a specific MetalConfig batched generate corruption bug; not the same fix." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "T5Tokenizer special-token method conflict is unrelated to custom model initialization in notebooks." + "reason": "Tokenizer special-method conflict and notebook custom-model initialization breakage are unrelated failures." }, { "left": "issue:44530", "right": "issue:46304", "accept": false, - "reason": "One is a Qwen3.5 cache/group-type crash; the other is a distributed-build compatibility regression. Different problems." + "reason": "PagedAttentionCache/Qwen3.5 crash vs distributed-build support regression; different code paths and symptoms." }, { - "left": "issue:40990", - "right": "issue:43421", + "left": "issue:44737", + "right": "issue:45859", "accept": false, - "reason": "High perplexity on gpt-oss-20b is not the same as runtime special-token post-processor desynchronization." + "reason": "XLNet CPU device placement bug and Qwen3.5 CPU offload incompatibility are unrelated model/runtime issues." }, { - "left": "issue:44737", - "right": "issue:45859", + "left": "issue:40990", + "right": "issue:43421", "accept": false, - "reason": "Both are device/performance-related, but one is CPU placement in XLNet positional encoding and the other is CPU offload incompatibility in Qwen3_5Moe rotary embedding." + "reason": "High perplexity on a model and runtime special-token post-processor updates are not the same underlying bug." }, { "left": "issue:40444", "right": "issue:46326", "accept": false, - "reason": "IterableDataset multi-image finetuning failure and mistral_common continue_final_message behavior are unrelated." + "reason": "Qwen2.5-VL multi-image iterable-dataset failure and mistral_common continue_final_message handling are distinct problems." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "TrainerState token accounting and Qwen2.5-VL multi-image dataset finetuning are separate issues." + "reason": "TrainerState token accounting and Qwen2.5-VL multi-image finetuning are unrelated issues." }, { "left": "issue:42175", "right": "issue:46178", "accept": false, - "reason": "Both involve backend packaging/mapping, but one is missing tensorflow in a torch extra and the other is an unmapped tensorflow_text backend; not the same bug." + "reason": "Backend packaging/install problem versus BACKENDS_MAPPING validation error; different failure modes." }, { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "Both are v5 initialization regressions, but one is non-persistent buffers filled with junk while the other is custom model initialization failing; distinct symptoms and likely distinct fixes." + "reason": "Both are Transformers v5 regressions around custom model/init behavior, but one is non-persistent buffer corruption and the other is custom model initialization failure; too different to treat as one bug." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "A template-node compilation TypeError is unrelated to GraniteMoEHybrid calling an invalid method." + "reason": "Template compilation TypeError and GraniteMoEHybrid invalid-method call are unrelated implementation bugs." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "AppleSilicon warmup TypeError and _is_hf_initialized flag regression are unrelated." - }, - { - "left": "issue:39404", - "right": "issue:44671", - "accept": false, - "reason": "Whisper return_language pipeline behavior and CamemBERT masked-LM prediction regression are different model families and bugs." + "reason": "Apple Silicon allocator TypeError and _is_hf_initialized flag breakage are different regressions." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "Both are audio-feature extraction related, but the specific failures are different and the titles do not indicate the same underlying code path." + "reason": "Both are audio-feature related, but one is a regression in LaserFeatureExtractor center handling and the other is a different fbank extraction issue; not clearly the same bug." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Separate tokenizer/model loading failures for different repositories; no evidence of a shared root cause." + "reason": "Both are transformers>=5 loading issues, but they target different models and different tokenizer/loading paths." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline return_language regression and CamemBERT masked-LM prediction regression are unrelated model behaviors." }, { "left": "issue:43452", "right": "issue:43482", "accept": false, - "reason": "Both involve GGUF loading, but one is a generic gguf_file break for AutoTokenizer/AutoModel loading and the other is a Qwen2.5-GGUF loading failure; too little to treat as the same bug." + "reason": "GGUF loading breaks in a generic from_pretrained path vs a specific Qwen2.5-GGUF v5 failure; related area, but not the same concrete bug." }, { "left": "issue:43482", "right": "issue:47312", "accept": false, - "reason": "Qwen2.5-GGUF loading failure and kernels 0.16 incompatibility are unrelated." + "reason": "A Qwen2.5-GGUF loading regression and a kernels 0.16 incompatibility do not appear to share the same code-path problem." }, { "left": "issue:44811", "right": "issue:45584", "accept": false, - "reason": "Whisper batch_decode skip_special_tokens handling and Whisper empty-transcription generation after align_special_tokens are related by model family but not the same concrete bug." + "reason": "Whisper batch_decode skip_special_tokens handling and empty-transcription generation after align_special_tokens are different bugs, though both are in Whisper." }, { - "left": "issue:43208", - "right": "issue:47013", + "left": "issue:43262", + "right": "issue:47331", "accept": false, - "reason": "Both are xLSTM issues, but one reports three training blockers for small models and the other is a specific return_last_states=False unpacking crash." + "reason": "Audio chat-template sampling-rate defaulting and MetalConfig quantization batch corruption are unrelated." }, { "left": "issue:43582", "right": "issue:47331", "accept": false, - "reason": "AppleSilicon warmup TypeError is unrelated to MetalConfig batched-generate corruption in quantization." + "reason": "Apple Silicon warmup TypeError and MetalConfig batched generate corruption are separate backend-specific issues." }, { - "left": "issue:43262", - "right": "issue:47331", + "left": "issue:43208", + "right": "issue:47013", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and MetalConfig quantization corruption are unrelated." + "reason": "Both are xLSTM issues, but one is a collection of training blockers and the other is a specific return_last_states=False unpacking crash; not one merged fix." }, { "left": "issue:42915", "right": "issue:45084", "accept": false, - "reason": "Qwen3Moe FP8 loading and template-node compilation errors are unrelated failures." + "reason": "Qwen3Moe FP8 loading failure and template compilation TypeError are unrelated." + }, + { + "left": "issue:43792", + "right": "issue:46869", + "accept": false, + "reason": "Whisper-large-v2 run failure and YOLO multi-GPU device-map auto failure are different models and different runtime paths." } ] }, @@ -3739,6 +3709,7 @@ "issue:28530", "issue:34567", "issue:36032", + "issue:40444", "issue:41720", "issue:42915", "issue:43089", @@ -3757,7 +3728,6 @@ "issue:43646", "issue:43742", "issue:43746", - "issue:43792", "issue:43819", "issue:43994", "issue:44008", @@ -3766,6 +3736,7 @@ "issue:44291", "issue:44403", "issue:44451", + "issue:44464", "issue:44466", "issue:44479", "issue:44534", @@ -3781,243 +3752,249 @@ "issue:46326", "issue:46661", "issue:46821", - "issue:46869", "issue:46899", "issue:47254" ], "soft_pairs": [ - "issue:43792|issue:46869", - "issue:43645|issue:46661", + "issue:44403|issue:45081", "issue:43746|issue:46326", + "issue:43645|issue:46661", "issue:43317|issue:43482", - "issue:44403|issue:45081", "issue:42915|issue:44291", - "issue:43646|issue:44534", "issue:44987|issue:45081", + "issue:43646|issue:44534", "issue:43262|issue:43377", "issue:36032|issue:43742", "issue:43377|issue:43819", "issue:42915|issue:44589", "issue:43994|issue:44610", - "issue:28530|issue:34567", "issue:43531|issue:44877", - "issue:44265|issue:44610", + "issue:28530|issue:34567", + "issue:44451|issue:45081", "issue:44062|issue:44987", "issue:43404|issue:44466", - "issue:44451|issue:45081", "issue:36032|issue:41720", + "issue:44265|issue:44610", "issue:44008|issue:46899", "issue:44849|issue:45325", "issue:44479|issue:45325", "issue:43632|issue:46821", - "issue:43089|issue:45987", - "issue:43322|issue:43540", "issue:43425|issue:44403", + "issue:43322|issue:43540", + "issue:43089|issue:45987", "issue:43296|issue:43531", - "issue:45923|issue:47254" + "issue:45923|issue:47254", + "issue:44464|issue:45987", + "issue:40444|issue:45325" ], - "prepared_review_unit_hash": "43000026749dc75bc0a830d3b0491b21ff63dffc0e0f24ac228eadf4dd7e068f", + "prepared_review_unit_hash": "06eead36d32b8044a8fba94c313cd3ae39b396d60d84a75e3ef6de8191e3cecc", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36560, - "estimated_input_tokens": 9140, - "estimated_eval_tokens": 18536 + "soft_pair_count": 30, + "serialized_chars": 36943, + "estimated_input_tokens": 9236, + "estimated_eval_tokens": 18728 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are mostly unrelated issue reports spanning different models, tokenizer paths, multimodal regressions, and infrastructure changes. I don\u2019t see any soft edge that clearly represents the same underlying bug/change, and there are no PRs in the set.", - "confidence": 0.96, - "canonical_issue_reason": "No clear duplicate cluster emerges from the titles alone; the reports cover distinct codepaths and model families. If a single representative is needed, issue 44479 is the broadest user-facing regression report.", + "summary": "This cluster is mostly a grab bag of unrelated Transformers bugs, with only loose thematic overlap around v5/init/load regressions. No soft pair looks like the same underlying bug strongly enough to merge, so all soft edges are rejected.", + "confidence": 0.68, + "canonical_issue_reason": "Issue 43632 is the closest thing to a broad umbrella bug here: it describes a concrete v5 initialization regression, which is the most reusable anchor among several nearby init/load-related reports.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44479 is the strongest representative because it is a broad, versioned regression affecting several Qwen multimodal variants and is clearly scoped as a user-facing bug.", + "best_issue_reason": "Issue 43632 is the best single issue anchor because it is concrete, broadly phrased, and closer to a root-cause regression than the more model-specific or symptom-only reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43792", - "right": "issue:46869", - "accept": false, - "reason": "Different models and failures: Whisper runtime issue vs YOLO multi-GPU device-map failure." - }, - { - "left": "issue:43645", - "right": "issue:46661", + "left": "issue:44403", + "right": "issue:45081", "accept": false, - "reason": "Both involve v5/custom-model loading, but they hit different codepaths: notebook initialization vs remote-code config_class lookup." + "reason": "Generic loading noise vs a tokenizer regex patch crash; different failure modes and code paths." }, { "left": "issue:43746", "right": "issue:46326", "accept": false, - "reason": "Unrelated: PEFT adapter checkpoint loading vs tokenizer continue_final_message behavior." + "reason": "PEFT local checkpoint loading and mistral_common continue_final_message are unrelated bugs." }, { - "left": "issue:43317", - "right": "issue:43482", + "left": "issue:43645", + "right": "issue:46661", "accept": false, - "reason": "Different subsystems and symptoms: gpu+cpu offload loading failure vs GGUF loading failure in v5." + "reason": "Both involve initialization, but one is notebook custom-model setup and the other is remote-code config_class resolution." }, { - "left": "issue:44403", - "right": "issue:45081", + "left": "issue:43317", + "right": "issue:43482", "accept": false, - "reason": "Noise on model loading is not the same bug as the Mistral regex patch crash." + "reason": "Dequantized model offload failure and Qwen2.5-GGUF loading failure are distinct model-loading problems." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "Both mention v5-era loading problems, but one is FP8 config handling and the other is init_empty_weights/_is_hf_initialized." + "reason": "Both touch v5/weights setup, but one is FineGrainedFP8Config on Qwen3Moe and the other is an init_empty_weights argument mismatch." }, { - "left": "issue:43646", - "right": "issue:44534", + "left": "issue:44987", + "right": "issue:45081", "accept": false, - "reason": "Custom model initialization breakage is distinct from non-persistent buffer corruption." + "reason": "A model load regression and a tokenizer regex crash are not the same underlying bug." }, { - "left": "issue:44987", - "right": "issue:45081", + "left": "issue:43646", + "right": "issue:44534", "accept": false, - "reason": "Different models and failure modes; no concrete shared codepath is evident." + "reason": "Custom model initialization and non-persistent buffer corruption are different issues." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Audio chat-template sampling-rate default is unrelated to MIMI encoder padding-mask batching." + "reason": "Audio chat-template sampling-rate handling and MIMI padding-mask batching are different modality bugs." }, { "left": "issue:36032", "right": "issue:43742", "accept": false, - "reason": "Tokenizer method-name conflict and MobileLLM key error are separate loading bugs." + "reason": "T5Tokenizer special-token method conflict and MobileLLM key error are unrelated." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "Different audio model bugs: padding-mask batching vs DAC latent/forward mismatch." + "reason": "MIMI padding-mask support and DAC from_latents STE mismatch are separate model-specific defects." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "FP8 config failure is unrelated to Float8 storage deserialization." + "reason": "Float8 config failure and missing Float8 storage registration are only loosely related by dtype." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Siglip2 bad outputs and OmDet-Turbo image-size mismatch are different multimodal issues." + "reason": "Siglip2 bad outputs and OmDet-Turbo processor size mismatch are different model/processor bugs." }, { - "left": "issue:28530", - "right": "issue:34567", + "left": "issue:43531", + "right": "issue:44877", "accept": false, - "reason": "Training early-stopping metric handling is unrelated to token-count tracking in TrainerState." + "reason": "Qwen3-MoE sliding_window behavior and granite_speech config strictness are unrelated." }, { - "left": "issue:43531", - "right": "issue:44877", + "left": "issue:28530", + "right": "issue:34567", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and granite_speech strict config loading are unrelated." + "reason": "Early stopping metric selection and num_input_tokens_seen updates are different trainer bookkeeping issues." }, { - "left": "issue:44265", - "right": "issue:44610", + "left": "issue:44451", + "right": "issue:45081", "accept": false, - "reason": "torch.export export failures and processor image-size mismatch are different bugs." + "reason": "A model loading failure and a tokenizer patch crash are not the same bug." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "AddedToken keyword duplication is unrelated to the physical-intelligence/fast loading failure." + "reason": "AddedToken keyword duplication and a model load regression have no concrete overlap." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Both touch tied weights, but one is a missing weight tie in a specific model and the other is serialization inconsistency across devices." - }, - { - "left": "issue:44451", - "right": "issue:45081", - "accept": false, - "reason": "ScandiBERT loading failure has no clear relation to the Mistral tokenizer regex crash." + "reason": "Tied lm_head weights at model definition time versus serialization inconsistency are related in theme but not the same code-path bug." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Tokenizer conflict and Qwen3 auto device-map GPU assert are unrelated." + "reason": "Tokenizer special-token conflict and Qwen3 auto device mapping cuda assert are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export/torch_compilable_check failure and OmDet input-size mismatch are separate problems." }, { "left": "issue:44008", "right": "issue:46899", "accept": false, - "reason": "Gemma3n variable collision and Gemma4 quantization dtype cast are different multimodal bugs." + "reason": "Both are multimodal, but one is a variable-name AttributeError and the other is a dtype-cast issue in embedders." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Qwen3.5 hidden-states bug and Qwen2.5-VL rope-index scaling bug affect different mechanisms." + "reason": "Different Qwen variants and different bugs: hidden-states output vs rope index scaling." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Related Qwen multimodal area, but 44479 is a broader video regression while 45325 is a specific still-image temporal-position scaling issue; not enough to call them the same bug." + "reason": "Video-input regression across Qwen models and still-image position-id scaling are not the same defect." }, { "left": "issue:43632", "right": "issue:46821", "accept": false, - "reason": "_is_hf_initialized flag breakage and TimesFM window_size attribute error are different failures." + "reason": "_is_hf_initialized initialization break and TimesFM window_size AttributeError are unrelated." }, { - "left": "issue:43089", - "right": "issue:45987", + "left": "issue:43425", + "right": "issue:44403", "accept": false, - "reason": "Generation performance overhead is unrelated to StaticCache sequence-length return type." + "reason": "Torch version incompatibility and noisy loading output are completely different issues." }, { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Llava Next loading segfault and Qwen3OmniMoe video processing ValueError are different model-path bugs." + "reason": "Both are multimodal loading bugs, but they involve different models and different concrete failures." }, { - "left": "issue:43425", - "right": "issue:44403", + "left": "issue:43089", + "right": "issue:45987", "accept": false, - "reason": "Torch version incompatibility is not the same as noisy model-loading logs." + "reason": "Generation performance overhead and StaticCache return-type mismatch are unrelated." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL loading in vLLM and Qwen3-MoE sliding-window behavior are unrelated." + "reason": "PaddleOCR-VL loading in vLLM and Qwen3-MoE sliding_window behavior are different bugs." }, { "left": "issue:45923", "right": "issue:47254", "accept": false, - "reason": "Missing supports_gradient_checkpointing on a trust_remote_code model is not the same as a PEFT+DeepSpeed checkpointing error." + "reason": "Missing supports_gradient_checkpointing and a PEFT/DeepSpeed checkpoint error are both about checkpointing, but not the same issue." + }, + { + "left": "issue:44464", + "right": "issue:45987", + "accept": false, + "reason": "Chunked-generation inconsistency with compiled forward and StaticCache.get_seq_length type mismatch are different code paths." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "IterableDataset multi-image finetuning failure and Qwen2.5-VL temporal position scaling are unrelated." } ] }, @@ -4050,6 +4027,7 @@ "issue:42796", "issue:43296", "issue:43381", + "issue:43421", "issue:43482", "issue:43525", "issue:43531", @@ -4062,26 +4040,26 @@ "issue:43883", "issue:44079", "issue:44220", + "issue:44265", "issue:44291", "issue:44387", "issue:44403", - "issue:44464", "issue:44488", "issue:44534", + "issue:44568", "issue:44589", "issue:44683", "issue:44756", "issue:44811", "issue:44991", + "issue:45072", "issue:45081", "issue:45216", - "issue:45325", "issue:45357", "issue:45542", "issue:45663", "issue:45812", "issue:45923", - "issue:45987", "issue:46315", "issue:46491", "issue:46710", @@ -4090,23 +4068,21 @@ "issue:47404" ], "soft_pairs": [ - "issue:44464|issue:45987", - "issue:40444|issue:45325", "issue:44220|issue:44683", "issue:43872|issue:44291", "issue:43716|issue:46899", "issue:43611|issue:44534", - "issue:44991|issue:46315", "issue:44403|issue:44991", - "issue:43883|issue:44488", - "issue:45216|issue:45357", + "issue:44991|issue:46315", "issue:44291|issue:44589", - "issue:42796|issue:43746", - "issue:45812|issue:46491", + "issue:45216|issue:45357", + "issue:43883|issue:44488", "issue:43792|issue:44589", + "issue:45812|issue:46491", + "issue:42796|issue:43746", "issue:33666|issue:40444", - "issue:43632|issue:44756", "issue:43381|issue:45923", + "issue:43632|issue:44756", "issue:40990|issue:44811", "issue:43531|issue:44291", "issue:35707|issue:40444", @@ -4114,223 +4090,232 @@ "issue:42796|issue:46710", "issue:41720|issue:45081", "issue:43482|issue:43525", - "issue:43883|issue:44991", - "issue:44387|issue:47404", "issue:43296|issue:46821", "issue:42175|issue:43296", - "issue:42175|issue:45542" + "issue:44387|issue:47404", + "issue:43883|issue:44991", + "issue:42175|issue:45542", + "issue:43421|issue:44568", + "issue:44079|issue:44534", + "issue:44265|issue:45072" ], - "prepared_review_unit_hash": "24ab88ff4771bbeab52a980064a646f1d52225a96810921e944d96447c27317d", + "prepared_review_unit_hash": "490733add68c62e0c34f53e1182a38e59822bef0708c681d3e37dc49114233af", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35984, - "estimated_input_tokens": 8996, - "estimated_eval_tokens": 18248 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36926, + "estimated_input_tokens": 9232, + "estimated_eval_tokens": 18720 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of unrelated issues. Two soft pairs look like true duplicates: the `_is_hf_initialized` loading regression pair and the Qwen3.5 `save_pretrained` checkpoint-key regression pair; the rest are too broad or topic-mismatched to merge.", - "confidence": 0.86, - "canonical_issue_reason": null, + "summary": "This cluster is mostly a loose similarity set of unrelated issues. It does contain two accepted duplicate/follow-up pairs: the `_is_hf_initialized` compatibility regression and the Qwen3.5 `save_pretrained` visual-key regression. The rest are different models, subsystems, or failure modes.", + "confidence": 0.88, + "canonical_issue_reason": "issue:44291 is the clearest, most general statement of the `_is_hf_initialized` loading regression and best captures the root cause for that duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:44291 is the strongest representative issue for the `_is_hf_initialized` regression in this cluster; it is the best anchor for that pair, though the cluster also includes a separate accepted Qwen3.5 `save_pretrained` pair.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44464", - "right": "issue:45987", - "accept": false, - "reason": "Different generation bugs: compiled-forward inconsistency vs `StaticCache.get_seq_length()` returning the wrong shape." - }, - { - "left": "issue:40444", - "right": "issue:45325", - "accept": false, - "reason": "Both are Qwen2.5-VL, but one is finetuning on iterable multi-image batches and the other is a RoPE/position-id scaling regression." - }, { "left": "issue:44220", "right": "issue:44683", "accept": false, - "reason": "Unrelated code paths: fbank extraction vs compiled flex_attention on torch 2.9+." + "reason": "Different bugs: one is about `_torch_extract_fbank_features()`, the other about `torch.export`/`flex_attention` on torch>=2.9." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same underlying `_is_hf_initialized` regression during model loading; 44291 is the broader report and 43872 is a bitsandbytes manifestation." + "reason": "Same underlying compatibility break around `_is_hf_initialized`; one report is bitsandbytes-specific, the other is the generic upstream regression." }, { "left": "issue:43716", "right": "issue:46899", "accept": false, - "reason": "Different multimodal dtype problems in different models; not the same bug or fix." + "reason": "Both are multimodal dtype problems, but they affect different models and code paths; not the same concrete bug." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "Separate Transformers v5 regressions with different failure modes: `base_model_prefix` loading vs junk-filled non-persistent buffers." + "reason": "Unrelated regressions: base_model_prefix loading vs non-persistent buffer initialization." }, { - "left": "issue:44991", - "right": "issue:46315", + "left": "issue:44403", + "right": "issue:44991", "accept": false, - "reason": "Tokenizer load failure vs performance regression in `convert_tokens_to_ids`; different symptoms and fixes." + "reason": "Different failure modes: generic load noise vs tokenizer loading failure for a specific model." }, { - "left": "issue:44403", - "right": "issue:44991", + "left": "issue:44991", + "right": "issue:46315", "accept": false, - "reason": "Loading-noise cleanup vs tokenizer loading breakage; not the same issue." + "reason": "Tokenizer loading failure is not the same as slower `convert_tokens_to_ids`; different code paths." }, { - "left": "issue:43883", - "right": "issue:44488", + "left": "issue:44291", + "right": "issue:44589", "accept": false, - "reason": "Missing `all_tied_weights_keys` on Molmo vs failure to load a specific model tokenizer; unrelated." + "reason": "`_is_hf_initialized` constructor incompatibility is unrelated to the float8 storage lookup error." }, { "left": "issue:45216", "right": "issue:45357", "accept": true, - "reason": "Same Qwen3.5 `save_pretrained` regression producing incorrect checkpoints; 45357 is the narrower visual-encoder-keys manifestation." + "reason": "Same Qwen3.5 `save_pretrained` regression; the second is a narrower follow-up about the incorrect visual encoder keys." }, { - "left": "issue:44291", - "right": "issue:44589", + "left": "issue:43883", + "right": "issue:44488", "accept": false, - "reason": "Both mention loading/runtime failures, but one is `_is_hf_initialized` and the other is Float8 storage lookup; different bugs." + "reason": "Missing `all_tied_weights_keys` on Molmo is unrelated to loading `cjvt/sleng-bert`." }, { - "left": "issue:42796", - "right": "issue:43746", + "left": "issue:43792", + "right": "issue:44589", "accept": false, - "reason": "Tokenizer BPE-marker handling vs GraniteSpeech PEFT checkpoint loading; unrelated." + "reason": "Whisper runtime failure and float8 storage lookup error are different bugs." }, { "left": "issue:45812", "right": "issue:46491", "accept": false, - "reason": "Two tokenizer regressions, but different models and different behaviors: wrong Granite token IDs vs CodeLlama leading-space round-trip." + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." }, { - "left": "issue:43792", - "right": "issue:44589", + "left": "issue:42796", + "right": "issue:43746", "accept": false, - "reason": "Whisper model runtime failure vs Float8 storage error; not the same code-path problem." + "reason": "Tokenizer BPE-marker handling is unrelated to GraniteSpeech PEFT local-checkpoint loading." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Both are Qwen VL training topics, but one is multi-GPU training and the other is iterable-dataset multi-image finetuning failure." + "reason": "Both are Qwen-VL training issues, but one is multi-GPU training and the other is iterable-dataset multi-image batching." }, { - "left": "issue:43632", - "right": "issue:44756", + "left": "issue:43381", + "right": "issue:45923", "accept": false, - "reason": "`_is_hf_initialized` regression vs mmap/OOM on Strix Halo; unrelated." + "reason": "Eval-mode gradient-checkpointing behavior is not the same bug as a missing `supports_gradient_checkpointing` flag." }, { - "left": "issue:43381", - "right": "issue:45923", + "left": "issue:43632", + "right": "issue:44756", "accept": false, - "reason": "Gradient-checkpointing eval-mode bug vs missing `supports_gradient_checkpointing` flag; related only loosely." + "reason": "Core v5 `_is_hf_initialized` breakage is unrelated to mmap OOM behavior on Strix Halo." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "Perplexity regression on GPT-OSS vs Whisper `batch_decode` skipping specials; unrelated." + "reason": "Perplexity regression and Whisper `batch_decode` special-token handling are unrelated." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Qwen3-MoE sliding-window bug vs `_is_hf_initialized` loading bug; different subsystems." + "reason": "Qwen3-MoE sliding-window behavior is unrelated to the `_is_hf_initialized` loading regression." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is unrelated to Qwen2.5-VL multi-image finetuning." + "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is a different bug from Qwen2.5-VL iterable-dataset failures." }, { "left": "issue:44079", "right": "issue:45663", "accept": false, - "reason": "ModelOutput key assignment bug vs Gemma4 FSDP2 shared-kv-state rebuild bug; unrelated." + "reason": "ModelOutput key assignment and Gemma4 FSDP2 shared-KV rebuild are different defects." }, { "left": "issue:42796", "right": "issue:46710", "accept": false, - "reason": "Tokenizer marker handling vs accuracy regression after upgrade; different symptoms and likely fixes." + "reason": "Tokenizer BPE-marker handling is unrelated to the DeepSeek accuracy regression." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Qwen3 device-mapping CUDA assert vs Mistral regex patch crash; unrelated." + "reason": "Auto device-map CUDA assert and Mistral regex patch crash are different code paths." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "GGUF loading failure with Transformers v5 vs missing `pad_token_id` on `Llama4Config`; different issues." + "reason": "GGUF loading failure and missing `pad_token_id` on Llama4Config are unrelated." }, { - "left": "issue:43883", - "right": "issue:44991", + "left": "issue:43296", + "right": "issue:46821", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and TimesFM `window_size` attribute errors are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43296", "accept": false, - "reason": "Model attribute missing vs tokenizer loading regression; no shared code-path." + "reason": "TensorFlow backend packaging and PaddleOCR-VL/vLLM loading are different issues." }, { "left": "issue:44387", "right": "issue:47404", "accept": false, - "reason": "Both concern OOMs, but one is int4 quantization reserved memory and the other is device-map transients during conversion; not the same bug." + "reason": "Both mention OOM, but one is int4 reserved-memory regression and the other is device_map conversion transients." }, { - "left": "issue:43296", - "right": "issue:46821", + "left": "issue:43883", + "right": "issue:44991", "accept": false, - "reason": "PaddleOCR-VL/vLLM load failure vs TimesFM `window_size` attribute error; unrelated." + "reason": "Molmo tied-weights attribute error is unrelated to tokenizer loading for EMBEDDIA/est-roberta." }, { "left": "issue:42175", - "right": "issue:43296", + "right": "issue:45542", "accept": false, - "reason": "TensorFlow backend packaging issue vs PaddleOCR-VL model loading failure; different layers entirely." + "reason": "Both touch TensorFlow availability, but one is missing TF in the torch extra and the other is a different backend-init problem when only tensorboard is present." }, { - "left": "issue:42175", - "right": "issue:45542", + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime special-token post-processor syncing is not the same as add_special_tokens failing for one tokenizer." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer corruption are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:45072", "accept": false, - "reason": "Both touch TensorFlow/backend packaging, but one is missing TF in the torch extra and the other is a runtime undefined-backend error from tensorboard-only installs." + "reason": "torch.export + torch_compilable_check is unrelated to bfloat16 dtype mismatches in inference." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary stays grounded in the packet and is suitably conservative. The two accepted soft pairs are well-supported by the titles as the same underlying regressions, while the rest are plausibly distinct and the rejections avoid overclaiming mergeability. No obvious overstatement or unsupported linkage." + "accept": false, + "feedback": "The cluster summary is mostly grounded, and the soft-edge rejections are conservative. However, the rationale for issue:44291 overstates the structure of the cluster: it says 44291 is the anchor for the \"only clear duplicate pair,\" but the packet also supports a second accepted duplicate/follow-up pair for Qwen3.5 save_pretrained (45216/45357). Please revise that wording so it only claims 44291 is the best anchor for the _is_hf_initialized pair, not the only duplicate pair in the cluster." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -4348,13 +4333,15 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ + "issue:34634", + "issue:35707", "issue:40444", "issue:40990", + "issue:41720", "issue:42796", "issue:43262", "issue:43317", "issue:43377", - "issue:43421", "issue:43454", "issue:43482", "issue:43606", @@ -4368,18 +4355,16 @@ "issue:43873", "issue:43994", "issue:44079", + "issue:44186", "issue:44265", "issue:44291", "issue:44462", "issue:44466", "issue:44479", "issue:44509", - "issue:44534", - "issue:44568", "issue:44829", "issue:44877", "issue:44898", - "issue:45072", "issue:45325", "issue:45357", "issue:45381", @@ -4398,11 +4383,8 @@ "issue:47217" ], "soft_pairs": [ - "issue:43421|issue:44568", - "issue:44079|issue:44534", - "issue:44265|issue:45072", - "issue:44462|issue:45812", "issue:43482|issue:45538", + "issue:44462|issue:45812", "issue:43611|issue:43646", "issue:45325|issue:45834", "issue:43317|issue:43856", @@ -4411,224 +4393,228 @@ "issue:40444|issue:45381", "issue:43644|issue:44079", "issue:43377|issue:45491", - "issue:43994|issue:44898", "issue:45325|issue:45381", + "issue:43994|issue:44898", "issue:40444|issue:47217", "issue:43454|issue:44466", - "issue:45325|issue:46697", "issue:43873|issue:45701", + "issue:45325|issue:46697", "issue:43606|issue:44265", - "issue:43818|issue:45357", "issue:43818|issue:46402", + "issue:43818|issue:45357", "issue:44479|issue:45823", "issue:42796|issue:45381", "issue:43828|issue:46821", "issue:40990|issue:45584", "issue:44829|issue:45479", - "issue:43262|issue:43746" + "issue:43262|issue:43746", + "issue:41720|issue:45823", + "issue:44186|issue:44265", + "issue:35707|issue:41720", + "issue:34634|issue:35707" ], - "prepared_review_unit_hash": "52040439af72fc75f019d4c0c2c5d0dea305af06aee57f290662addee569a218", + "prepared_review_unit_hash": "892b22bc5e289e95ee78f993df9ff3cac033263f8d6ca02c32d28a3d4fcbe935", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36482, - "estimated_input_tokens": 9121, - "estimated_eval_tokens": 18498 + "soft_pair_count": 29, + "serialized_chars": 36675, + "estimated_input_tokens": 9169, + "estimated_eval_tokens": 18594 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated issues with no clear duplicate cluster. The only near-match is the Qwen2.5-VL position-id regression pair, but the titles still point to different symptoms/code paths, so I rejected all soft edges.", - "confidence": 0.66, - "canonical_issue_reason": "Issue 45325 is the most concrete and code-path-specific report in the only semi-related subset (Qwen2.5-VL positional/vision index regressions).", + "summary": "This cluster is only loosely related: most items are separate Transformers v5 regressions across tokenization, model loading, multimodal inputs, device mapping, and docs. None of the soft edges look like true duplicates of the same underlying bug or change.", + "confidence": 0.23, + "canonical_issue_reason": "Issue 43611 is the broadest representative of the cluster\u2019s recurring v5 model-loading breakages; several other items are adjacent to loading/init/serialization regressions, but the cluster is not a single duplicate bug.", "canonical_pr_reason": null, - "best_issue_reason": "45325 is the best representative issue here because it names the exact failing function and root symptom, making it the clearest triage anchor.", + "best_issue_reason": "Issue 43611 is the best stand-in for the cluster because it is the most generic and central of the v5 loading regressions, though it does not actually subsume most of the other issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43421", - "right": "issue:44568", - "accept": false, - "reason": "Both are tokenizer-related, but one is runtime post-processor syncing and the other is BOS/EOS insertion; different bugs." - }, - { - "left": "issue:44079", - "right": "issue:44534", - "accept": false, - "reason": "ModelOutput key assignment and non-persistent buffer corruption are unrelated code paths." - }, - { - "left": "issue:44265", - "right": "issue:45072", + "left": "issue:43482", + "right": "issue:45538", "accept": false, - "reason": "torch.export/torch_compilable_check failure is a different problem from bf16 dtype mismatches in inference." + "reason": "Both are tokenizer/loading-related, but one is a GGUF loading failure and the other is a CLIPTokenizer max-length bug; different code paths and symptoms." }, { "left": "issue:44462", "right": "issue:45812", "accept": false, - "reason": "Both involve tokenizers, but one ignores tokenizer.json while the other returns wrong Granite token IDs; not the same bug." - }, - { - "left": "issue:43482", - "right": "issue:45538", - "accept": false, - "reason": "GGUF loading failure and CLIPTokenizer max_length behavior are unrelated." + "reason": "Both involve AutoTokenizer behavior, but one ignores tokenizer.json while the other returns wrong token IDs for Granite models; not the same bug." }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Base_model_prefix loading breakage and custom model initialization breakage are different regressions." + "reason": "Both are v5 loading regressions, but one is about base_model_prefix loading and the other about custom model initialization; distinct failures." }, { "left": "issue:45325", "right": "issue:45834", "accept": false, - "reason": "Qwen2.5-VL rope indexing and Kosmos2.5 OCR long-input indexing are separate model-specific issues." + "reason": "Both are multimodal indexing bugs, but one is Qwen2.5-VL rope scaling for still images and the other is Kosmos2.5 OCR index overflow; different issues." }, { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "Device-map offload loading failure is not the same as Qwen3 MoE training memory inefficiency." + "reason": "One is a device-map/offload loading failure, the other is Qwen3 MoE training memory inefficiency; not the same underlying bug." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "Unexpected _is_hf_initialized argument and strict config loading are different load-path bugs." + "reason": "Both affect loading/config initialization, but one is an init_empty_weights argument error and the other is strict config rejection for granite_speech; separate root causes." }, { "left": "issue:44509", "right": "issue:46909", "accept": false, - "reason": "Both are documentation issues, but they concern different docs and different examples/tasks." + "reason": "Both are documentation-related, but they cover unrelated docs problems (removed pipeline tasks vs a broken DIA example)." }, { "left": "issue:40444", - "right": "issue:47217", + "right": "issue:45381", "accept": false, - "reason": "Multi-image IterableDataset finetuning failure is not the same as image-placeholder-in-user-text inference failure." + "reason": "Both are Qwen2.5-VL multimodal bugs, but one is finetuning with multiple images and the other is video vision_position_ids; different paths." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "Filling non-persistent buffers with junk and misassigned ModelOutput keys are unrelated." + "reason": "One concerns non-persistent buffer serialization, the other ModelOutput key assignment; unrelated internals." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "Missing padding-mask support in MIMI batching and Gemma3 NaNs from sliding-window padding are different bugs." + "reason": "Both involve batching/padding side effects, but they are different models and different failures: missing padding-mask support vs NaN embeddings from sliding-window attention." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL positional-id issues, but one is still-image temporal scaling and the other is video input vision_position_ids; not the same concrete bug." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "SigLIP AutoModel/pipeline nonsense outputs and Perceiver interpolation failures are separate model behaviors." + "reason": "Different models and bugs: SigLIP2 nonsense outputs with AutoModel/pipeline versus Perceiver non-default-resolution failure." }, { - "left": "issue:43818", - "right": "issue:45357", + "left": "issue:40444", + "right": "issue:47217", "accept": false, - "reason": "Video-LLaVA temporal-attention/weight-sharing bug and Qwen3.5 visual-key save_pretrained regression are unrelated." + "reason": "Both mention image placement in VL prompts, but one is finetuning with multiple images per prompt and the other is user-text placeholder handling; not clearly the same bug." }, { - "left": "issue:43828", - "right": "issue:46821", + "left": "issue:43454", + "right": "issue:44466", "accept": false, - "reason": "Autocast dtype mismatch and TimesFM window_size attribute error do not share a concrete code path." + "reason": "Both touch lm_head weights, but one is missing weight tying in a config and the other is inconsistent save_pretrained serialization depending on device." }, { - "left": "issue:40990", - "right": "issue:45584", + "left": "issue:43873", + "right": "issue:45701", "accept": false, - "reason": "High perplexity on gpt-oss and Whisper empty-transcription failure are different symptoms and models." + "reason": "One is a quantization/offloading runtime issue and the other is a tokenizer behavior change; unrelated." }, { - "left": "issue:44829", - "right": "issue:45823", + "left": "issue:45325", + "right": "issue:46697", "accept": false, - "reason": "Flash-attention training degeneration and Gemma4 device-mismatch forward failure are not the same bug." + "reason": "Different subsystems entirely: Qwen2.5-VL rope indexing versus Swin2SR image preprocessing padding." }, { - "left": "issue:42796", - "right": "issue:45381", + "left": "issue:43606", + "right": "issue:44265", "accept": false, - "reason": "Tokenizer BPE-marker handling and Qwen2.5-VL video position-id regression are separate issues." + "reason": "Bark CPU-offload device mismatch and torch.export failing on torch_compilable_check are unrelated problems." }, { - "left": "issue:45325", - "right": "issue:45381", + "left": "issue:43818", + "right": "issue:46402", "accept": false, - "reason": "Both are Qwen2.5-VL positional issues, but one is still-image temporal scaling and the other is video vision_position_ids; not clearly one concrete bug." + "reason": "Different models and failures: Video-LLaVA temporal-attention/weight-sharing bug versus Sam3Model text encoder weights not loading." }, { - "left": "issue:40444", - "right": "issue:45381", + "left": "issue:43818", + "right": "issue:45357", "accept": false, - "reason": "IterableDataset multi-image finetuning failure is unrelated to Qwen2.5-VL video position indexing." + "reason": "Video-LLaVA architecture bug and Qwen3.5 save_pretrained key regression are unrelated." }, { - "left": "issue:43454", - "right": "issue:44466", + "left": "issue:44479", + "right": "issue:45823", "accept": false, - "reason": "Tied lm_head weights/generation garbage and device-dependent serialization of lm_head.weight are related in theme but not the same failure." + "reason": "Both are device/runtime regressions, but one is a Qwen video-input regression and the other is a Gemma4 PLE device mismatch during forward." }, { - "left": "issue:45325", - "right": "issue:46697", + "left": "issue:42796", + "right": "issue:45381", "accept": false, - "reason": "Qwen2.5-VL rope index scaling and Swin2SR over-padding are unrelated image/video preprocessing bugs." + "reason": "Tokenizer BPE marker handling and Qwen2.5-VL video vision_position_ids are unrelated bugs." }, { - "left": "issue:43873", - "right": "issue:45701", + "left": "issue:43828", + "right": "issue:46821", "accept": false, - "reason": "Quantization/offloading behavior and tokenization changes are different problem areas." + "reason": "Different failure modes in different models: autocast dtype mismatch versus a missing window_size attribute." }, { - "left": "issue:43606", - "right": "issue:44265", + "left": "issue:40990", + "right": "issue:45584", "accept": false, - "reason": "CPU offload device mismatch and export failure with torch_compilable_check are separate code paths." + "reason": "High perplexity on WikiText-2 and Whisper empty-transcription failure are unrelated." }, { - "left": "issue:43818", - "right": "issue:46402", + "left": "issue:44829", + "right": "issue:45479", "accept": false, - "reason": "Video-LLaVA architectural bug and Sam3Model text-encoder loading issue are unrelated." + "reason": "Both are degenerate training/loss complaints, but one is flash_attention_3 behavior and the other is single-label classification with num_labels=1; different code paths." }, { - "left": "issue:44479", + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio processor chat-template sample-rate defaulting and GraniteSpeech PEFT checkpoint loading are separate issues." + }, + { + "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Qwen video regression and Gemma4 PLE device mismatch are different model families and bugs." + "reason": "Both involve device_map/forward placement, but they affect different models and distinct errors." }, { - "left": "issue:44509", - "right": "issue:46909", + "left": "issue:44186", + "right": "issue:44265", "accept": false, - "reason": "Documentation cleanup for removed pipelines and a broken DIA example are distinct docs issues." + "reason": "Tokenizer crashes on NER/padding and torch.export failures on torch_compilable_check are unrelated." }, { - "left": "issue:43262", - "right": "issue:43746", + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device-mapping cuda asserts." + }, + { + "left": "issue:34634", + "right": "issue:35707", "accept": false, - "reason": "Audio chat-template sample-rate defaulting and PEFT adapter checkpoint loading are unrelated." + "reason": "Bark voice_preset not working and progressive generation inputs_embeds/past_key_values are different bugs with no shared underlying code path." } ] }, @@ -4652,8 +4638,6 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ - "issue:34634", - "issue:35707", "issue:36331", "issue:40990", "issue:41553", @@ -4677,17 +4661,18 @@ "issue:43819", "issue:43873", "issue:44186", - "issue:44265", "issue:44387", "issue:44534", "issue:44568", "issue:44610", + "issue:44756", "issue:44877", "issue:44898", "issue:44977", "issue:44991", "issue:45072", "issue:45325", + "issue:45685", "issue:45701", "issue:45704", "issue:45759", @@ -4702,13 +4687,9 @@ "issue:47405" ], "soft_pairs": [ - "issue:44186|issue:44265", - "issue:41720|issue:45823", - "issue:35707|issue:41720", - "issue:34634|issue:35707", "issue:43606|issue:43819", - "issue:45823|issue:46821", "issue:42617|issue:42915", + "issue:45823|issue:46821", "issue:43454|issue:43550", "issue:40990|issue:42947", "issue:43257|issue:45923", @@ -4716,245 +4697,242 @@ "issue:47404|issue:47405", "issue:44568|issue:45920", "issue:43377|issue:44610", - "issue:42796|issue:46209", "issue:36331|issue:42617", - "issue:36331|issue:42915", + "issue:42796|issue:46209", "issue:44877|issue:44991", - "issue:41720|issue:43322", + "issue:36331|issue:42915", "issue:43454|issue:43720", + "issue:41720|issue:43322", "issue:43749|issue:44534", - "issue:43377|issue:44898", "issue:41720|issue:42617", + "issue:43377|issue:44898", "issue:43317|issue:43873", "issue:41720|issue:44977", - "issue:45325|issue:46315", "issue:43572|issue:45759", - "issue:41553|issue:43582", + "issue:45325|issue:46315", "issue:44387|issue:45704", - "issue:43322|issue:46050", "issue:40990|issue:45701", - "issue:43450|issue:45701" + "issue:43322|issue:46050", + "issue:43450|issue:45701", + "issue:41553|issue:43582", + "issue:45685|issue:46821", + "issue:43873|issue:44756", + "issue:44186|issue:45072" ], - "prepared_review_unit_hash": "2d7bc46a93a010cba2de02bef3d0f273ff0ef31e20f123303bd3abce9abe2923", + "prepared_review_unit_hash": "019053931ebc3676e4dc36aa41f510a0e70971ec3bbef35c35c49b008708d49a", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37342, - "estimated_input_tokens": 9336, - "estimated_eval_tokens": 18928 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36435, + "estimated_input_tokens": 9109, + "estimated_eval_tokens": 18474 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is a grab bag of distinct Transformers bug reports across tokenization, device mapping/offload, config loading, quantization, and model-specific failures. None of the soft pairs look like the same underlying bug or change, so there is no true duplicate cluster here.", - "confidence": 0.89, - "canonical_issue_reason": "No canonical issue: the items do not converge on one concrete bug, and the soft-pair candidates are all different failure modes or different model/code paths.", + "summary": "This is a heterogeneous cluster of mostly unrelated issues across tokenizer regressions, model-loading/offloading, quantization, and model-specific runtime bugs. I do not see any soft pair that is clearly the same underlying defect, so all soft links should be rejected. If a representative issue is needed, 43873 is the broadest offloading/quantization report, but it does not explain the whole cluster.", + "confidence": 0.9, + "canonical_issue_reason": "43873 is the broadest representative of the offloading/quantization subset and is the most central active loading-related report, but the cluster overall is too mixed to have a perfect canonical issue.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good global representative; the set is too heterogeneous to anchor on one report without collapsing unrelated regressions into one cluster.", + "best_issue_reason": "43873 is the most generally useful issue in the set: open, broad, and closest to the repeated offloading/device_map/quantization failures seen here.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44186", - "right": "issue:44265", - "accept": false, - "reason": "Different bugs: tokenizer NER/padding/truncation crash vs torch.export failure around torch_compilable_check." - }, - { - "left": "issue:41720", - "right": "issue:45823", - "accept": false, - "reason": "Both involve device placement, but they target different models and different concrete failures (cuda assert vs PLE mismatch)." - }, - { - "left": "issue:35707", - "right": "issue:41720", - "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device mapping cuda assert." - }, - { - "left": "issue:34634", - "right": "issue:35707", - "accept": false, - "reason": "Bark voice_preset support and generation cache behavior are different code paths and symptoms." - }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "Different model-path failures: CPU offload device mismatch vs DAC latent/forward mismatch from missing STE." + "reason": "Different failures: CPU-offload device mismatch for Bark vs DAC latent/STE mismatch in an audio codec path." }, { - "left": "issue:45823", - "right": "issue:46821", + "left": "issue:42617", + "right": "issue:42915", "accept": false, - "reason": "No shared underlying bug; Gemma4 device mismatch and TimesFM window_size AttributeError are unrelated." + "reason": "One is a 3d_parallel.py runtime problem; the other is a Qwen3Moe FineGrainedFP8Config failure." }, { - "left": "issue:42617", - "right": "issue:42915", + "left": "issue:45823", + "right": "issue:46821", "accept": false, - "reason": "3d_parallel.py runtime failure and Qwen3Moe FineGrainedFP8Config failure are separate issues." + "reason": "Gemma4 PLE device mismatch is unrelated to TimesFM's missing window_size attribute." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "AyaVision lm_head tying bug and Bamba torch.compile SDPA failure are different concrete problems." + "reason": "AyaVision lm_head weight tying and Bamba torch.compile+SDPA are separate code paths." }, { "left": "issue:40990", "right": "issue:42947", "accept": false, - "reason": "High perplexity on a model/dataset is not the same bug as gradient checkpointing being ineffective with LoRA." + "reason": "GPT-OSS perplexity regression is not the same bug as LoRA gradient-checkpointing ineffectiveness." }, { "left": "issue:43257", "right": "issue:45923", "accept": false, - "reason": "Different topics: MOE weight conversion under accelerate+deepspeed vs missing supports_gradient_checkpointing flag." + "reason": "Accelerate/deepspeed MOE weight conversion differs from a missing gradient_checkpointing flag." }, { "left": "issue:43454", "right": "issue:45072", "accept": false, - "reason": "Weight-tying garbage text and dtype mismatch in bfloat16 inference are unrelated failure modes." + "reason": "AyaVision weight tying is unrelated to bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:47404", "right": "issue:47405", "accept": false, - "reason": "Both concern loading/conversion, but one is OOM from transient ops and the other is silent wrong-class loading; not the same bug." + "reason": "Quantized loading OOM from destination-GPU transients is distinct from silent wrong-Auto-class conversion skip." }, { "left": "issue:44568", "right": "issue:45920", "accept": false, - "reason": "Specific add_special_tokens regression for one tokenizer is not the same as broad wrong-token-ID regressions across many models." + "reason": "Specific add_special_tokens BOS/EOS regression is different from broader wrong-token-id regressions." }, { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "Missing padding-mask support in MIMI encoder and batched video processor shape errors are different issues." - }, - { - "left": "issue:42796", - "right": "issue:46209", - "accept": false, - "reason": "Tokenizer BPE marker handling and FP8 inference on B200 are unrelated." + "reason": "MIMI padding-mask batching bug is unrelated to OmDet-Turbo's image-size mismatch." }, { "left": "issue:36331", "right": "issue:42617", "accept": false, - "reason": "CustomTrainer compute_loss signature breakage is unrelated to 3d_parallel.py failure." + "reason": "Trainer compute_loss signature mismatch and 3d_parallel.py failure are unrelated." }, { - "left": "issue:36331", - "right": "issue:42915", + "left": "issue:42796", + "right": "issue:46209", "accept": false, - "reason": "Trainer API kwarg mismatch and Qwen3Moe FP8 loading are different bugs." + "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are different bugs." }, { "left": "issue:44877", "right": "issue:44991", "accept": false, - "reason": "Config strictness for granite_speech and tokenizer loading regression for est-roberta are different code paths." + "reason": "Strict config rejection for granite_speech and tokenizer loading failure for est-roberta are different load paths." }, { - "left": "issue:41720", - "right": "issue:43322", + "left": "issue:36331", + "right": "issue:42915", "accept": false, - "reason": "Qwen3 device-mapping cuda assert and Llava Next segfault are different model-specific failures." + "reason": "Trainer API keyword mismatch and Qwen3Moe FP8 failure are unrelated." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "AyaVision weight tying and BitNet packed-weight unpacking are not the same underlying problem." + "reason": "AyaVision lm_head tying bug is not the same as BitNet packed-weight unpacking during accelerate loading." }, { - "left": "issue:43749", - "right": "issue:44534", + "left": "issue:41720", + "right": "issue:43322", "accept": false, - "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk values are unrelated regressions." + "reason": "Qwen3 auto device-map cuda assert and Llava Next segfault are different model/load failures." }, { - "left": "issue:43377", - "right": "issue:44898", + "left": "issue:43749", + "right": "issue:44534", "accept": false, - "reason": "MIMI batching/padding-mask bug and Perceiver resolution/interpolation issue are different model behaviors." + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are separate regressions." }, { "left": "issue:41720", "right": "issue:42617", "accept": false, - "reason": "Both are runtime/model loading failures, but they are unrelated code paths and symptoms." + "reason": "Qwen3 auto device mapping assert is unrelated to 3d_parallel.py failing to run." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "MIMI batched padding-mask bug and Perceiver non-default resolution failure are unrelated." }, { "left": "issue:43317", "right": "issue:43873", "accept": false, - "reason": "Quantized dequantized-model loading on offload and generic offloading problems with quantization are related thematically but not the same concrete bug." + "reason": "Both involve offloading/quantization, but one is a dequantized-model load failure and the other is a broader offloading complaint; not clearly the same bug." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "Auto device mapping cuda assert for Qwen3 is unrelated to flash-attention generation issues in Qwen3.5." - }, - { - "left": "issue:45325", - "right": "issue:46315", - "accept": false, - "reason": "RoPE position-id scaling and convert_tokens_to_ids slowdown are separate tokenizer/model issues." + "reason": "Auto device-map crash and flash-attention generation issue in Qwen3.5 are different symptoms/code paths." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig and unwrapping text_config for composite Qwen configs are different config-loading bugs." + "reason": "StableLm pad_token_idx regression and Qwen multimodal text_config unwrapping are unrelated config bugs." }, { - "left": "issue:41553", - "right": "issue:43582", + "left": "issue:45325", + "right": "issue:46315", "accept": false, - "reason": "Tokenizer load error messaging and Apple Silicon caching_allocator_warmup TypeError are unrelated." + "reason": "Qwen2.5-VL rope-index scaling and slow convert_tokens_to_ids are different tokenizer/model behaviors." }, { "left": "issue:44387", "right": "issue:45704", "accept": false, - "reason": "Int4 quantization memory regression and apex FusedRMSNorm leak are different performance/memory bugs." + "reason": "int4 quantization OOM from reserved memory is not the same as T5's apex RMSNorm memory leak." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "GPT-OSS perplexity regression and tokenizer-version tokenization changes are different issues." }, { "left": "issue:43322", "right": "issue:46050", "accept": false, - "reason": "Llava Next segfault on load and sinq save/load corruption are different failure modes." + "reason": "Llava Next loading segfault and sinq save/load breakage are unrelated quantized-model bugs." }, { - "left": "issue:40990", + "left": "issue:43450", "right": "issue:45701", "accept": false, - "reason": "High perplexity on one model/dataset is not the same as a version-wide tokenization change." + "reason": "Video processor batch-shape bug and tokenizer-version tokenization change do not match." }, { - "left": "issue:43450", - "right": "issue:45701", + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Bad AutoTokenizer error message for Voxtral is unrelated to Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:45685", + "right": "issue:46821", + "accept": false, + "reason": "MPS histogram on Int and TimesFM window_size attribute error are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Strix Halo mmap OOM is a hardware-specific mmap issue, not the same as quantization offloading regression." + }, + { + "left": "issue:44186", + "right": "issue:45072", "accept": false, - "reason": "Batched video processor shape bug and tokenizer behavior change are unrelated." + "reason": "LayoutLMv2 tokenizer crash and bfloat16 dtype mismatch are unrelated." } ] }, @@ -4980,6 +4958,7 @@ "nodes": [ "issue:26413", "issue:28530", + "issue:30064", "issue:33666", "issue:41720", "issue:42915", @@ -4987,6 +4966,7 @@ "issue:43295", "issue:43317", "issue:43404", + "issue:43441", "issue:43531", "issue:43550", "issue:43572", @@ -4996,22 +4976,19 @@ "issue:43708", "issue:43716", "issue:43720", - "issue:43873", "issue:43881", "issue:44008", "issue:44186", + "issue:44361", "issue:44423", "issue:44451", "issue:44466", "issue:44488", "issue:44610", - "issue:44756", "issue:44898", - "issue:45072", "issue:45161", "issue:45357", "issue:45406", - "issue:45685", "issue:45701", "issue:45704", "issue:45823", @@ -5025,234 +5002,235 @@ "issue:46897", "issue:47020", "issue:47317", + "issue:47337", "issue:47404" ], "soft_pairs": [ - "issue:45685|issue:46821", - "issue:43873|issue:44756", - "issue:44186|issue:45072", "issue:44466|issue:45357", - "issue:26413|issue:28530", "issue:43295|issue:45701", + "issue:26413|issue:28530", "issue:44451|issue:47020", - "issue:43550|issue:43720", "issue:43065|issue:46402", + "issue:43550|issue:43720", "issue:43550|issue:43606", "issue:43577|issue:43708", + "issue:46821|issue:46869", "issue:44186|issue:44610", "issue:43646|issue:46665", - "issue:46821|issue:46869", "issue:46612|issue:46858", "issue:41720|issue:45161", "issue:46897|issue:47317", "issue:44008|issue:45823", - "issue:44423|issue:45406", "issue:42915|issue:43317", + "issue:44423|issue:45406", "issue:44610|issue:44898", "issue:28530|issue:33666", "issue:42915|issue:46291", "issue:43404|issue:45357", "issue:43531|issue:43572", "issue:45704|issue:47404", + "issue:43716|issue:46821", + "issue:28530|issue:30064", "issue:43881|issue:44488", - "issue:43716|issue:46821" + "issue:43441|issue:47337", + "issue:44186|issue:44361" ], - "prepared_review_unit_hash": "32f355ef4286f1ea0e5c287863a9ebfce629ed3be795fdaf45226e0a583be4d7", + "prepared_review_unit_hash": "3f127c80bc7bc9728727b0766640ba74556052e6665316e04a6a56b976f96270", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36201, - "estimated_input_tokens": 9051, - "estimated_eval_tokens": 18358 + "serialized_chars": 36200, + "estimated_input_tokens": 9050, + "estimated_eval_tokens": 18356 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated Transformers bug reports that only share broad vocabulary like device_map, loading, tokenizer, dtype, or offloading. None of the soft pairs look like the same underlying bug/change, so I would not merge any of them as duplicates.", - "confidence": 0.94, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: the soft pairs mostly share only broad subsystems (tokenizers, loading, multimodal, device_map, training), not the same concrete bug. I would not merge any of the suggested duplicate edges.", + "confidence": 0.92, + "canonical_issue_reason": "No single issue in this set cleanly represents a duplicate cluster; the items cover unrelated bugs across training, loading, tokenization, multimodal processors, and device mapping.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the strongest representative by activity and breadth of description, but it still does not match the rest of the set closely enough to serve as a true duplicate canonical.", + "best_issue_reason": "No best issue candidate is convincing because the cluster does not appear to describe one underlying problem.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:45685", - "right": "issue:46821", - "accept": false, - "reason": "Different failures: MPS int histogram support vs TimesFM window_size attribute handling." - }, - { - "left": "issue:43873", - "right": "issue:44756", - "accept": false, - "reason": "Both involve offloading/memory, but the root causes differ: quantization offload behavior vs mmap OOM on Strix Halo." - }, - { - "left": "issue:44186", - "right": "issue:45072", - "accept": false, - "reason": "Unrelated model/tokenizer crash vs bfloat16 dtype mismatch in different wrappers." - }, { "left": "issue:44466", "right": "issue:45357", "accept": false, - "reason": "Both concern serialization, but one is tied lm_head persistence and the other is incorrect visual encoder keys in Qwen3.5." + "reason": "Both mention save/serialization, but one is tied lm_head serialization and the other is incorrect visual encoder keys; different bugs and code paths." }, { - "left": "issue:26413", - "right": "issue:28530", + "left": "issue:43295", + "right": "issue:45701", "accept": false, - "reason": "Trainer bugs in different subsystems: checkpoint resume iterator failure vs early stopping metric lookup." + "reason": "One is a processor.tokenizer regression in v4.57.5; the other is a broader tokenization change. Not the same failure." }, { - "left": "issue:43295", - "right": "issue:45701", + "left": "issue:26413", + "right": "issue:28530", "accept": false, - "reason": "Both are version-related regressions, but they affect different APIs and code paths." + "reason": "Resume-from-checkpoint epoch iterator failure and early-stopping metric lookup are unrelated trainer issues." }, { "left": "issue:44451", "right": "issue:47020", "accept": false, - "reason": "Model loading failure for two different causes: one package/model compatibility issue vs Unigram expecting a sequence." + "reason": "Both are model-loading/tokenizer-related, but they point to different models and different root causes." }, { - "left": "issue:43550", - "right": "issue:43720", + "left": "issue:43065", + "right": "issue:46402", "accept": false, - "reason": "Different problems: torch.compile/SDPA on Bamba vs accelerate loading of packed BitNet weights." + "reason": "Both concern Sam3, but one is a dummy Conv2d in the pixel decoder while the other is missing text-encoder weight loading." }, { - "left": "issue:43065", - "right": "issue:46402", + "left": "issue:43550", + "right": "issue:43720", "accept": false, - "reason": "Same model family, but dummy Conv2d in the pixel decoder is not the same bug as missing Sam3 text encoder weights." + "reason": "Different models and different failure modes: torch.compile with SDPA vs accelerate loading of packed weights." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Both are device/runtime issues, but they target different models and distinct failure mechanisms." + "reason": "Both are device/runtime bugs, but one is SDPA compilation and the other is CPU offload device mismatch." }, { "left": "issue:43577", "right": "issue:43708", "accept": false, - "reason": "Dtype loading bug vs Trainer max_steps miscalculation are unrelated." + "reason": "Blip2 dtype propagation and Trainer max_steps on checkpoint resume are unrelated." + }, + { + "left": "issue:46821", + "right": "issue:46869", + "accept": false, + "reason": "TimesFM window_size AttributeError and YOLO multi-GPU device_map auto are different subsystems and bugs." }, { "left": "issue:44186", "right": "issue:44610", "accept": false, - "reason": "Both are processor/model mismatches, but they are different models and different incorrect inputs." + "reason": "Tokenizer crash on NER/padding is unrelated to OmDet-Turbo image-size mismatch." }, { "left": "issue:43646", "right": "issue:46665", "accept": false, - "reason": "Custom model initialization regression is not the same as inconsistent from_pretrained behavior inside Python classes." - }, - { - "left": "issue:46821", - "right": "issue:46869", - "accept": false, - "reason": "TimesFM window_size attribute error is unrelated to YOLO multi-GPU device-map loading." + "reason": "Both involve model initialization/loading, but one is a Transformers 5.0 breakage in custom init and the other is inconsistent from_pretrained behavior in Python classes." }, { "left": "issue:46612", "right": "issue:46858", "accept": false, - "reason": "Both touch cache logic, but one is skipped beam-search reorder in specific models and the other is a static-cache slice-arithmetic bug on non-CUDA backends." + "reason": "Both touch generation/cache, but one is skipped beam-search cache reorder and the other is a static-cache seq_length device-tensor arithmetic bug." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "Different distributed-loading/parallelism failures on different models and code paths." + "reason": "Different models and different parallelism problems: auto device_map CUDA assert vs tensor-parallel-only failure." }, { "left": "issue:46897", "right": "issue:47317", "accept": false, - "reason": "Both are training-loss bugs, but they are model-specific and do not appear to share the same underlying defect." + "reason": "Same general symptom class (training loss), but different models and no evidence of the same underlying bug." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Gemma3n variable collision in forward is unrelated to Gemma4 PLE device mismatch." + "reason": "Different Gemma models and different bugs: variable-name collision in forward vs PLE device mismatch." }, { - "left": "issue:44423", - "right": "issue:45406", + "left": "issue:42915", + "right": "issue:43317", "accept": false, - "reason": "Both crash in serve with multimodal models, but the missing attributes and affected processors differ." + "reason": "FineGrainedFP8Config failure and dequantized-model offload failure are distinct loading/configuration issues." }, { - "left": "issue:42915", - "right": "issue:43317", + "left": "issue:44423", + "right": "issue:45406", "accept": false, - "reason": "Quantization/FineGrainedFP8 failure is a different loading path from dequantized model offload on gpu+cpu." + "reason": "Both affect serve/multimodal paths, but they fail in different places with different processor/model assumptions." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Both are image-size/preprocessing mismatches, but they concern different architectures and different expected resolutions." + "reason": "Both are vision preprocessing issues, but one is a fixed-size processor/model mismatch and the other is non-default-resolution behavior with interpolate_pos_encoding." }, { "left": "issue:28530", "right": "issue:33666", "accept": false, - "reason": "Early stopping metric detection and Qwen2-VL multi-GPU training are unrelated." + "reason": "Early stopping metric resolution and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:42915", "right": "issue:46291", "accept": false, - "reason": "Quantization loading failure is unrelated to pipeline import failing on kernels version/revision requirements." + "reason": "FineGrainedFP8 loading and kernels import-version requirements are unrelated." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "Tied lm_head weights in Mistral3 is a different bug from incorrect visual encoder keys in Qwen3.5 save_pretrained." + "reason": "Mistral3 tied-weight serialization and Qwen3.5 visual encoder key serialization are different bugs." }, { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and StableLmConfig missing pad_token_idx are unrelated." + "reason": "Sliding-window behavior in Qwen3-MoE and missing pad_token_idx in StableLmConfig are unrelated model/config issues." }, { "left": "issue:45704", "right": "issue:47404", "accept": false, - "reason": "Different memory problems: apex RMSNorm leak vs loading OOM from destination-GPU contiguous conversion work." + "reason": "Apex RMSNorm memory leak and device_map loading OOM from conversion transients are different problems." + }, + { + "left": "issue:43716", + "right": "issue:46821", + "accept": false, + "reason": "Image preprocessor/model dtype mismatch and TimesFM window_size AttributeError do not share a code path." + }, + { + "left": "issue:28530", + "right": "issue:30064", + "accept": false, + "reason": "Trainer early stopping and image processor void-segmentation handling are unrelated." }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "Both are model loading failures, but they are distinct models with no shared concrete root cause." + "reason": "Both are loading failures, but for different models with no shared root cause shown." }, { - "left": "issue:43716", - "right": "issue:46821", + "left": "issue:43441", + "right": "issue:47337", "accept": false, - "reason": "Mistral-3 image preprocessor dtype mismatch is unrelated to TimesFM window_size attribute errors." + "reason": "FlashAttention failure for Ministral-3 and a flaky bbox-range test are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Different tokenizers, different tasks, and different AttributeError causes." } ] }, @@ -5276,19 +5254,21 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ - "issue:28530", "issue:29159", + "issue:29925", "issue:30064", "issue:33357", + "issue:34567", "issue:34634", "issue:35707", "issue:36032", "issue:41720", + "issue:42490", "issue:42915", "issue:43257", "issue:43316", "issue:43322", - "issue:43441", + "issue:43329", "issue:43452", "issue:43531", "issue:43540", @@ -5299,17 +5279,16 @@ "issue:43746", "issue:43782", "issue:43873", - "issue:43881", - "issue:44186", - "issue:44361", "issue:44387", "issue:44464", "issue:44534", + "issue:44589", "issue:44671", "issue:44734", "issue:44743", "issue:44913", - "issue:44991", + "issue:45406", + "issue:45464", "issue:45479", "issue:45632", "issue:45684", @@ -5322,228 +5301,234 @@ "issue:46894", "issue:47020", "issue:47254", - "issue:47317", - "issue:47337" + "issue:47317" ], "soft_pairs": [ - "issue:28530|issue:30064", - "issue:43441|issue:47337", - "issue:44186|issue:44361", "issue:44734|issue:46858", "issue:44464|issue:46858", "issue:30064|issue:34634", - "issue:29159|issue:36032", "issue:43452|issue:47020", + "issue:29159|issue:36032", "issue:44671|issue:44743", - "issue:42915|issue:43322", "issue:44387|issue:45701", + "issue:42915|issue:43322", "issue:43873|issue:45704", - "issue:43540|issue:43782", + "issue:43531|issue:45759", "issue:45479|issue:47317", "issue:33357|issue:35707", - "issue:43531|issue:45759", + "issue:43540|issue:43782", "issue:43550|issue:43653", "issue:43316|issue:44913", "issue:41720|issue:43746", - "issue:44534|issue:46665", "issue:45632|issue:45684", + "issue:44534|issue:46665", "issue:43606|issue:43720", - "issue:46894|issue:47317", "issue:43257|issue:47254", - "issue:43881|issue:44991", + "issue:46894|issue:47317", "issue:43531|issue:46821", - "issue:34634|issue:41720" + "issue:34634|issue:41720", + "issue:43531|issue:44589", + "issue:42490|issue:43329", + "issue:29159|issue:29925", + "issue:45406|issue:45464", + "issue:34567|issue:35707" ], - "prepared_review_unit_hash": "1c44237fc4d0856881652225b616863750a1cba350a7d9848511ba364e3f2946", + "prepared_review_unit_hash": "5fc7ae71dcb6eb429a4f3182f0fdbc984455aec3fe159deaf02a022aeb9047f6", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 27, - "serialized_chars": 35730, - "estimated_input_tokens": 8933, - "estimated_eval_tokens": 18122 + "soft_pair_count": 28, + "serialized_chars": 36132, + "estimated_input_tokens": 9033, + "estimated_eval_tokens": 18322 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is very heterogeneous: most pairs only share a broad theme like tokenizer/loading/generation/offload bugs, but they do not describe the same concrete code-path failure. I would not merge any of the soft edges, and there is no clear canonical issue or PR representative.", - "confidence": 0.95, - "canonical_issue_reason": "No single issue is a credible canonical representative because the set spans unrelated bugs across early stopping, tokenizers, multimodal loading, generation, quantization, configuration, and training-loss paths.", - "canonical_pr_reason": "No PRs are present in the cluster.", - "best_issue_reason": "No global best issue stands out; the only overlaps are broad subsystem similarities, not the same underlying defect.", - "best_pr_reason": "No PRs are present in the cluster.", + "summary": "This cluster is highly heterogeneous: the soft pairs mostly share broad subsystem words (tokenizer, loading, cache, Qwen, windowing) but not the same concrete bug or change. I would not merge any of the candidate pairs.", + "confidence": 0.94, + "canonical_issue_reason": "issue:29159 is the broadest and most representative issue in the set, centered on slow-vs-fast tokenizer behavioral inconsistency, which is the closest to the few tokenizer-related candidates.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:29159 is the best single issue to keep as a representative anchor because it has the most general problem statement and the widest semantic overlap with other tokenizer issues in the cluster.", + "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:28530", - "right": "issue:30064", - "accept": false, - "reason": "Early stopping metric selection vs image processor void segmentation maps are unrelated failures." - }, - { - "left": "issue:43441", - "right": "issue:47337", - "accept": false, - "reason": "FlashAttention model loading on MacOS and a flaky bbox-range test are different problems." - }, - { - "left": "issue:44186", - "right": "issue:44361", - "accept": false, - "reason": "Both are tokenizer crashes, but they affect different models and different failure modes." - }, { "left": "issue:44734", "right": "issue:46858", "accept": false, - "reason": "Both mention cache/indexing, but one is serve KV continuation and the other is static-cache slice arithmetic on non-CUDA backends." + "reason": "Both involve cache/compiled-generation failures, but one is KV-cache continuation indexing in serve and the other is static-cache sequence-length arithmetic on non-CUDA backends." }, { "left": "issue:44464", "right": "issue:46858", "accept": false, - "reason": "Chunked generation with compiled forward is not the same bug as device-tensor seq_length in prefill." + "reason": "Both concern generation under compile/cache, but the bugs are different: chunked generation output mismatch vs device-tensor seq-length breaking slice arithmetic." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Image segmentation-map processing and BarkProcessor voice_preset handling are unrelated." + "reason": "Unrelated bugs in different areas: void segmentation map processing vs Bark voice_preset handling." }, { - "left": "issue:29159", - "right": "issue:36032", + "left": "issue:43452", + "right": "issue:47020", "accept": false, - "reason": "A slow/fast tokenizer inconsistency is not the same as a T5 special-token method conflict on load." + "reason": "Both are tokenizer-loading related, but one is gguf_file/from_pretrained path handling and the other is a constructor type mismatch in XLMRobertaTokenizer." }, { - "left": "issue:43452", - "right": "issue:47020", + "left": "issue:29159", + "right": "issue:36032", "accept": false, - "reason": "Generic gguf loading breakage and a specific XLMRobertaTokenizer/Unigram constructor mismatch are different code paths." + "reason": "Both are tokenizer issues, but one is slow/fast behavioral inconsistency while the other is a T5 class/method name conflict during loading." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "CamemBERT masked LM predictions and Qwen3 recurrent-state reset under cache are unrelated." + "reason": "Different code paths and symptoms: incorrect CamemBERT masked-LM outputs vs recurrent state reset when caching with seq_len>1." }, { - "left": "issue:42915", - "right": "issue:43322", + "left": "issue:44387", + "right": "issue:45701", "accept": false, - "reason": "FineGrainedFP8Config failure and Llava Next segfault are different loading/runtime bugs." + "reason": "Int4 quantization memory growth/OOM is unrelated to the tokenization changes described in the other issue." }, { - "left": "issue:44387", - "right": "issue:45701", + "left": "issue:42915", + "right": "issue:43322", "accept": false, - "reason": "Int4 CUDA memory growth is not the same as a tokenizer-version behavior change." + "reason": "Both are model-load failures, but one is Qwen3Moe fine-grained FP8 config handling and the other is a Llava Next segmentation fault." }, { "left": "issue:43873", "right": "issue:45704", "accept": false, - "reason": "Quantization offloading behavior and T5/apex RMSNorm memory leak are unrelated." + "reason": "Different bugs: quantization/offloading behavior vs a T5 apex RMSNorm memory leak." }, { - "left": "issue:43540", - "right": "issue:43782", + "left": "issue:43531", + "right": "issue:45759", "accept": false, - "reason": "Qwen3OmniMoE video processing failure and Qwen3VL weight_only loading error are different issues." + "reason": "Both mention Qwen3, but one is sliding_window behavior in Qwen3-MoE and the other is config unwrapping for composite multimodal models." }, { "left": "issue:45479", "right": "issue:47317", "accept": false, - "reason": "A generic sequence-classification zero-loss bug is not enough to match a separate PPFormulaNet training-loss report." + "reason": "Both are training-loss problems, but there is no evidence they share the same loss computation bug or code path." }, { "left": "issue:33357", "right": "issue:35707", "accept": false, - "reason": "CLIP MacOS bus error and progressive generation with inputs_embeds/past_key_values are different bugs." + "reason": "A MacOS CLIP bus error is unrelated to progressive generation with inputs_embeds/past_key_values." }, { - "left": "issue:43531", - "right": "issue:45759", + "left": "issue:43540", + "right": "issue:43782", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and composite Qwen config unwrapping are different defects." + "reason": "Both are Qwen multimodal issues, but one is a video-input processing ValueError and the other is a from_pretrained weight_only load error." }, { "left": "issue:43550", "right": "issue:43653", "accept": false, - "reason": "torch.compile/SDPA model failure and tokenizer special-token registration are unrelated." + "reason": "Completely different subsystems: torch.compile/SDPA in Bamba vs BigBird tokenizer special-token registration." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Config API inconsistency and rotary_pct reload defaulting are separate configuration bugs." + "reason": "Both involve config behavior, but one is an API discrepancy in Gemma3TextConfig and the other is rotary_pct not persisting on reload." }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Qwen3 auto device-mapping CUDA assert and GraniteSpeech PEFT checkpoint loading are different failure modes." - }, - { - "left": "issue:44534", - "right": "issue:46665", - "accept": false, - "reason": "Non-persistent buffer initialization and from_pretrained-in-classes behavior are not the same bug." + "reason": "Different problems: device-mapping cudaErrorAssert on Qwen3 vs GraniteSpeech PEFT checkpoint loading." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Cache-path collisions and propagated read-only permissions are different filesystem issues." + "reason": "One is a trust_remote_code cache-path collision and the other is save_pretrained permission propagation; same area, different bug." }, { - "left": "issue:43606", - "right": "issue:43720", + "left": "issue:44534", + "right": "issue:46665", "accept": false, - "reason": "CPU offload device mismatch and packed-weight unpacking during accelerate loading are separate problems." + "reason": "Non-persistent buffer initialization in v5 is unrelated to from_pretrained behavior inside Python classes." }, { - "left": "issue:46894", - "right": "issue:47317", + "left": "issue:43606", + "right": "issue:43720", "accept": false, - "reason": "The titles suggest different model-specific training-loss bugs; the shared phrase is too broad to merge." + "reason": "Both involve loading/offload/accelerate, but one is a device mismatch with CPU offload and the other is missing unpacking of packed BitNet weights." }, { "left": "issue:43257", "right": "issue:47254", "accept": false, - "reason": "Qwen3 MoE weight conversion during accelerate+deepspeed loading and ZeRO-3 checkpoint errors are different code paths." + "reason": "Both touch distributed training/loading, but the concrete failures are different: Qwen3 MOE weight conversion vs PEFT ZeRO-3 checkpointing." }, { - "left": "issue:43881", - "right": "issue:44991", + "left": "issue:46894", + "right": "issue:47317", "accept": false, - "reason": "Generic model loading failure and tokenizer loading regression are not the same issue." + "reason": "Both are training-loss bugs, but they refer to different models and no shared implementation defect is evident." }, { "left": "issue:43531", "right": "issue:46821", "accept": false, - "reason": "Qwen3-MoE sliding_window and TimesFM window_size AttributeError are unrelated despite both mentioning window/window_size." + "reason": "Window-related names aside, sliding_window in Qwen3-MoE and TimesFM window_size attribute errors are different bugs." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "BarkProcessor voice_preset and Qwen3 device mapping/CUDA assert are unrelated." + "reason": "BarkProcessor voice_preset handling and Qwen3 auto device mapping failures are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "No shared bug: Qwen3-MoE sliding_window behavior vs Float8 storage lookup failure." + }, + { + "left": "issue:42490", + "right": "issue:43329", + "accept": false, + "reason": "Local-path trailing-slash import warning is unrelated to the multimodal video-token counting bug." + }, + { + "left": "issue:29159", + "right": "issue:29925", + "accept": false, + "reason": "Related theme, but not the same bug: generic slow/fast tokenizer inconsistency vs SigLIP specifically not forcing use_fast=True." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Different runtime failures in serve/chat APIs: missing _tokenizer on Gemma4Processor vs Qwen3.5 streaming inference crash." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token-count tracking and progressive generation with inputs_embeds/past_key_values are unrelated issues." } ] }, @@ -5567,33 +5552,33 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ - "issue:29159", - "issue:29925", "issue:30064", - "issue:34567", - "issue:35707", "issue:36010", "issue:39692", - "issue:42490", - "issue:43329", + "issue:42886", "issue:43388", + "issue:43421", "issue:43526", - "issue:43531", "issue:43540", "issue:43606", "issue:43644", "issue:43653", "issue:43746", "issue:43873", + "issue:43881", "issue:43883", + "issue:43927", "issue:44062", "issue:44077", "issue:44451", + "issue:44464", "issue:44492", "issue:44509", - "issue:44589", + "issue:44514", + "issue:44545", "issue:44610", "issue:44625", + "issue:44843", "issue:44898", "issue:44918", "issue:44936", @@ -5603,8 +5588,7 @@ "issue:45059", "issue:45072", "issue:45357", - "issue:45406", - "issue:45464", + "issue:45479", "issue:45704", "issue:45759", "issue:45859", @@ -5613,221 +5597,232 @@ "issue:46402", "issue:46665", "issue:46858", + "issue:47050", "issue:47333", "issue:47405" ], "soft_pairs": [ - "issue:29159|issue:29925", - "issue:42490|issue:43329", - "issue:43531|issue:44589", - "issue:45406|issue:45464", - "issue:34567|issue:35707", + "issue:43881|issue:44991", "issue:43606|issue:43653", "issue:44492|issue:44509", "issue:45759|issue:45859", - "issue:43526|issue:43883", - "issue:43540|issue:46858", "issue:44451|issue:44991", + "issue:43540|issue:46858", "issue:45704|issue:47405", + "issue:43526|issue:43883", "issue:43388|issue:45059", "issue:44987|issue:45042", - "issue:43644|issue:46665", "issue:44077|issue:44625", + "issue:43644|issue:46665", "issue:45923|issue:47333", "issue:44898|issue:45072", "issue:45357|issue:46402", - "issue:39692|issue:44610", "issue:44062|issue:44451", - "issue:44918|issue:44936", + "issue:39692|issue:44610", "issue:43746|issue:46050", "issue:30064|issue:36010", - "issue:43746|issue:43873" + "issue:44918|issue:44936", + "issue:43746|issue:43873", + "issue:42886|issue:44843", + "issue:43421|issue:43927", + "issue:44625|issue:45479", + "issue:44514|issue:44545", + "issue:44464|issue:47050" ], - "prepared_review_unit_hash": "bfe8f8d7fa2b29a48e418a39ab8daf0d18f1f9ec3b314c73b45f0bb744c417be", + "prepared_review_unit_hash": "2561c5b3e93d65e133264d2ab099ed734192930cbd7fc35674386eb4a55e2b7b", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 25, - "serialized_chars": 35398, - "estimated_input_tokens": 8850, - "estimated_eval_tokens": 17956 + "soft_pair_count": 26, + "serialized_chars": 35841, + "estimated_input_tokens": 8961, + "estimated_eval_tokens": 18178 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous. The only clearly related items are the tokenizer inconsistency reports, but none of the soft pairs look like true duplicates. No PRs are present.", - "confidence": 0.83, - "canonical_issue_reason": "Issue 29159 is the broadest, most general report in the tokenizer subcluster, so it is the best canonical candidate there.", + "summary": "Mostly unrelated issue set, with two plausible duplicate pairs: an exact duplicate Qwen2_5_VLProcessor crash report, and a likely shared HF_HUB_OFFLINE tokenizer-cache regression. The remaining soft edges are only broad topical similarity, not the same concrete bug.", + "confidence": 0.79, + "canonical_issue_reason": "Issue 44514 is the cleanest issue anchor: it is the earlier report in an exact duplicate pair with the same title as 44545, and it contains enough reproduction detail to represent that bug well.", "canonical_pr_reason": null, - "best_issue_reason": "As a cluster-wide representative it is still weak, because most other issues concern unrelated bugs; 29159 only captures the tokenizer inconsistency theme.", + "best_issue_reason": "Issue 44514 is the strongest issue to keep as the representative bug report because it exactly matches 44545 and cleanly describes the batched apply_chat_template crash.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:29159", - "right": "issue:29925", - "accept": false, - "reason": "Related tokenizer behavior, but 29159 is a broad slow-vs-fast inconsistency report while 29925 is a specific SigLIP use_fast enforcement bug." - }, - { - "left": "issue:42490", - "right": "issue:43329", - "accept": false, - "reason": "Different code paths: local path trailing-slash import warning vs multimodal token counting video-branch undefined variables." - }, - { - "left": "issue:43531", - "right": "issue:44589", - "accept": false, - "reason": "Unrelated failures: Qwen3-MoE sliding_window behavior vs Float8 storage import/type error." - }, - { - "left": "issue:45406", - "right": "issue:45464", - "accept": false, - "reason": "Both are inference/load failures, but one is Gemma4Processor missing _tokenizer in serve and the other is Qwen3.5 streaming chat/completions behavior." - }, - { - "left": "issue:34567", - "right": "issue:35707", + "left": "issue:43881", + "right": "issue:44991", "accept": false, - "reason": "Trainer state token counting not updating is distinct from progressive generation with inputs_embeds and past_key_values." + "reason": "Both are model/tokenizer loading failures, but they concern different models and different failure modes; too broad to treat as the same bug." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "CPU offload device mismatch for bark-small is unrelated to BigBirdTokenizer special-token decode behavior." + "reason": "Completely different subsystems: CPU offload device mismatch vs tokenizer special-token decoding behavior." }, { "left": "issue:44492", "right": "issue:44509", "accept": false, - "reason": "A typo in cache strategy text is not the same bug as docs still mentioning removed pipeline tasks." + "reason": "Unrelated: a cache-strategy typo versus documentation still mentioning removed pipeline tasks." }, { "left": "issue:45759", "right": "issue:45859", "accept": false, - "reason": "Composite config unwrapping in from_config is a different issue from Qwen3_5MoeTextRotaryEmbedding CPU offload compatibility." + "reason": "Both involve Qwen 3.5/3.6, but one is a from_config/text_config unwrap issue and the other is a CPU-offload rotary embedding bug; different code paths." }, { - "left": "issue:43526", - "right": "issue:43883", + "left": "issue:44451", + "right": "issue:44991", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug and Molmo missing all_tied_weights_keys are unrelated." + "reason": "Both mention loading failures, but the models and likely root causes differ; not the same concrete tokenizer bug." }, { "left": "issue:43540", "right": "issue:46858", "accept": false, - "reason": "Qwen3OmniMoe video-input ValueError and static-cache seq_length device-tensor arithmetic are separate bugs." + "reason": "Different features and failure surfaces: video input processing vs cache sequence length / torch.compile arithmetic." }, { - "left": "issue:44451", - "right": "issue:44991", + "left": "issue:45704", + "right": "issue:47405", "accept": false, - "reason": "Both are loading problems, but they refer to different models and likely different root causes." + "reason": "Unrelated problems: Apex RMSNorm memory leak in T5 versus wrong model class loading silently producing random weights." }, { - "left": "issue:45704", - "right": "issue:47405", + "left": "issue:43526", + "right": "issue:43883", "accept": false, - "reason": "T5 apex RMSNorm memory leak and silent wrong-Auto-class conversion skip are not the same concrete code path." + "reason": "Different areas entirely: BeitImageProcessorFast label reduction vs Molmo tied-weights attribute error." }, { "left": "issue:43388", "right": "issue:45059", "accept": false, - "reason": "Metric gathering dropping tuple labels is distinct from SAM3 PCS text/box behavior." + "reason": "Both involve vision/multimodal pipelines, but the symptoms and code paths are unrelated." }, { "left": "issue:44987", "right": "issue:45042", "accept": false, - "reason": "Model loading failure for physical-intelligence/fast is unrelated to PIL image processors requiring torchvision." + "reason": "Different regressions: a specific model load failure versus PIL image processors incorrectly requiring torchvision." }, { - "left": "issue:43644", - "right": "issue:46665", + "left": "issue:44077", + "right": "issue:44625", "accept": false, - "reason": "Non-persistent junk buffer initialization is a different bug from inconsistent from_pretrained behavior inside Python classes." + "reason": "Not the same bug; one is a patchtsmixer post_init policy issue, the other is Qwen3.5 num_labels propagation." }, { - "left": "issue:44077", - "right": "issue:44625", + "left": "issue:43644", + "right": "issue:46665", "accept": false, - "reason": "Optional post_init handling in patchtsmixer is unrelated to Qwen3.5 num_labels propagation." + "reason": "Both involve from_pretrained behavior, but one is about junk non-persistent buffers and the other is a Python-class usage inconsistency." }, { "left": "issue:45923", "right": "issue:47333", "accept": false, - "reason": "Missing gradient checkpointing support and disk-offload save failure are different offloading/model lifecycle bugs." + "reason": "Different features and failures: missing gradient_checkpointing support versus disk-offload save failure." }, { "left": "issue:44898", "right": "issue:45072", "accept": false, - "reason": "Perceiver non-default-resolution failure and bf16 dtype mismatch are separate runtime problems." + "reason": "Different model families and failure modes: interpolation at non-default resolution vs bfloat16 dtype mismatch." }, { "left": "issue:45357", "right": "issue:46402", "accept": false, - "reason": "Incorrect visual-encoder keys on save_pretrained is not the same as Sam3Model text-encoder weights not loading." - }, - { - "left": "issue:39692", - "right": "issue:44610", - "accept": false, - "reason": "Documentation example errors for SigLIP2 are unrelated to OmDet-Turbo processor/model input-size mismatch." + "reason": "Both are model save/load issues, but they affect different models and opposite directions (saving incorrect keys vs not loading weights)." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "AddedToken special-argument duplication is a tokenizer constructor bug, not a model loading issue." + "reason": "Tokenizer-related, but the specific AddedToken constructor error is not the same as a generic latest-version load failure." }, { - "left": "issue:44918", - "right": "issue:44936", + "left": "issue:39692", + "right": "issue:44610", "accept": false, - "reason": "TRL SFT embedding unpacking failure is unrelated to trainer.evaluate() failing after trainer.train()." + "reason": "Both mention processor/model mismatch, but one is a SigLIP2 docs/example issue and the other is an OmDet-Turbo image-size mismatch." }, { "left": "issue:43746", "right": "issue:46050", "accept": false, - "reason": "PEFT local-checkpoint loading for GraniteSpeech is a different problem from sinq quantized model save/load breakage." + "reason": "Different persistence bugs: PEFT adapter loading from local checkpoints versus quantized model save/load corruption." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Void segmentation-map processing failure and GenerationMixin import error are unrelated subsystems." + "reason": "Completely unrelated: segmentation-map processing versus an import error for GenerationMixin." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "Both touch trainers, but they describe different failure points and likely different root causes." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "Both mention loading/offloading, but one is PEFT local checkpoint loading and the other is quantization offloading not behaving as expected." + "reason": "Not the same bug; one is PEFT checkpoint loading, the other is a general offloading/quantization behavior issue." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": true, + "reason": "These describe the same HF_HUB_OFFLINE tokenizer-cache regression; 44843 provides the likely root cause in _patch_mistral_regex/model_info." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Different tokenizer/config bugs: runtime special-token post-processing versus DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "Different classification bugs with different scopes and symptoms; not the same underlying issue." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Exact same title and same batched apply_chat_template crash; these are duplicate reports of the same bug." + }, + { + "left": "issue:44464", + "right": "issue:47050", + "accept": false, + "reason": "Both mention generation/inference, but one is chunked generation with compiled forward and the other is an ONNX export dtype mismatch." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the accepted pairs are conservative and well-supported by the titles: 44514/44545 are an exact duplicate pair, and 42886/44843 plausibly describe the same offline tokenizer-cache bug. The remaining soft-edge rejections stay appropriately cautious about different models, subsystems, or failure modes." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -5844,15 +5839,18 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ + "issue:30064", + "issue:36331", "issue:39692", - "issue:42886", "issue:43295", "issue:43381", "issue:43421", "issue:43493", + "issue:43531", "issue:43540", "issue:43550", "issue:43596", + "issue:43643", "issue:43653", "issue:43673", "issue:43720", @@ -5862,16 +5860,14 @@ "issue:43873", "issue:43927", "issue:43994", + "issue:44062", "issue:44186", "issue:44220", "issue:44368", - "issue:44464", - "issue:44514", - "issue:44545", - "issue:44625", + "issue:44488", + "issue:44661", "issue:44671", "issue:44821", - "issue:44843", "issue:44857", "issue:44995", "issue:45072", @@ -5880,7 +5876,6 @@ "issue:45356", "issue:45357", "issue:45381", - "issue:45479", "issue:45834", "issue:45907", "issue:46032", @@ -5889,253 +5884,257 @@ "issue:46655", "issue:46693", "issue:46858", + "issue:47020", "issue:47050", "issue:47337" ], "soft_pairs": [ - "issue:43421|issue:43927", - "issue:42886|issue:44843", - "issue:44625|issue:45479", - "issue:44514|issue:44545", "issue:46655|issue:46693", - "issue:44464|issue:47050", "issue:44995|issue:46858", "issue:39692|issue:43493", "issue:43381|issue:43844", "issue:43421|issue:43653", "issue:43673|issue:47337", "issue:43782|issue:45381", - "issue:45081|issue:45356", "issue:44857|issue:47050", + "issue:45081|issue:45356", "issue:46032|issue:46463", "issue:44368|issue:45357", - "issue:43421|issue:44186", "issue:39692|issue:47050", - "issue:43994|issue:44671", + "issue:43421|issue:44186", "issue:39692|issue:43873", + "issue:43994|issue:44671", "issue:45356|issue:45907", "issue:43720|issue:43819", - "issue:43540|issue:46294", "issue:43596|issue:44220", + "issue:43540|issue:46294", "issue:43550|issue:43819", "issue:43550|issue:43927", "issue:43295|issue:45834", "issue:44821|issue:45084", "issue:45072|issue:47050", - "issue:43653|issue:43720" + "issue:30064|issue:36331", + "issue:43653|issue:43720", + "issue:44062|issue:44488", + "issue:44488|issue:47020", + "issue:44220|issue:44661", + "issue:43531|issue:43643" ], - "prepared_review_unit_hash": "5956b412e3c9d988ff2f8c46538a7d60b726627eee95651783cfd32a01bddc31", + "prepared_review_unit_hash": "145ff995c5f50678ad77afdc4794d4f15bec804e37951ae3485d98c5c66c00ce", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36531, - "estimated_input_tokens": 9133, - "estimated_eval_tokens": 18522 + "serialized_chars": 36992, + "estimated_input_tokens": 9248, + "estimated_eval_tokens": 18752 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "One clear near-duplicate pair stands out: the Qwen2_5_VLProcessor.apply_chat_template batched padding=False crash reports. Most other soft edges are better treated as same-subsystem but distinct bugs; in particular, the offline-tokenizer pair is related but not grounded enough from titles alone to merge as the same root cause.", - "confidence": 0.76, - "canonical_issue_reason": "issue:44545 is the cleanest canonical anchor for the near-duplicate Qwen2_5_VLProcessor.apply_chat_template batched crash reports; its title is slightly clearer and it has more inbound references.", + "summary": "No soft pair looks like a true duplicate; these are mostly unrelated issues across different models, tokenizer paths, generation, training, and documentation. The cluster does not have a clear canonical issue or PR.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue: the items span distinct subsystems and failure modes, and none of the soft pairs appears to describe the same underlying bug.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44545 is the best representative issue in this set because it is the clearest duplicate-ready report and the most concise description of the underlying crash.", + "best_issue_reason": "No global best issue: there is no cohesive duplicate cluster here, so any one issue would be an arbitrary representative rather than a true canonical.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43421", - "right": "issue:43927", + "left": "issue:46655", + "right": "issue:46693", "accept": false, - "reason": "Different bugs: runtime special-token/post-processor sync vs save/load losing custom token IDs and causing generation IndexError." - }, - { - "left": "issue:42886", - "right": "issue:44843", - "accept": false, - "reason": "Related offline-tokenizer symptoms, but the titles point to different concrete failure modes; the packet does not establish the same root cause." - }, - { - "left": "issue:44625", - "right": "issue:45479", - "accept": false, - "reason": "Both are sequence-classification/config issues, but one is missing num_labels propagation and the other is a degenerate zero-loss case with num_labels=1 across models." - }, - { - "left": "issue:44514", - "right": "issue:44545", - "accept": true, - "reason": "Same Qwen2_5_VLProcessor.apply_chat_template batched padding=False crash, described twice with only minor wording differences." - }, - { - "left": "issue:46655", - "right": "issue:46693", - "accept": false, - "reason": "Video padding bug vs long-context FA2 performance regression; different code paths and failure modes." - }, - { - "left": "issue:44464", - "right": "issue:47050", - "accept": false, - "reason": "Chunked-generation inconsistency with compiled forward is unrelated to the Canine fp16 ONNX attention dtype mismatch." + "reason": "Video padding/calculation bug vs long-context FA2 performance regression; different code paths and symptoms." }, { "left": "issue:44995", "right": "issue:46858", "accept": false, - "reason": "Both involve cache logic, but one is a stale indexer cache on second forward and the other is device-tensor seq_length arithmetic in static cache/compile." + "reason": "Stale indexer cache on second forward vs cache length tensor type in static-cache torch.compile; unrelated failures." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "SigLIP2 docs/example errors are not the same as the HF-vs-JAX implementation discrepancy." + "reason": "SigLIP2 documentation/example errors vs HF/JAX implementation discrepancy; one is usage docs, the other is model behavior." }, { "left": "issue:43381", "right": "issue:43844", "accept": false, - "reason": "Eval-mode gradient checkpointing restriction is unrelated to the ZeRO-3 gradient growth regression." + "reason": "Eval-mode checkpointing restriction vs ZeRO-3 gradient inflation during training; not the same bug." }, { "left": "issue:43421", "right": "issue:43653", "accept": false, - "reason": "Runtime post-processor updates for special tokens are a different bug than BigBirdTokenizer's mask token registration problem." + "reason": "Runtime special-token/post-processor update bug vs BigBirdTokenizer mask-token registration bug; different tokenizer mechanics." }, { "left": "issue:43673", "right": "issue:47337", "accept": false, - "reason": "Generation cache missing in chunked prefill is unrelated to the flaky bbox coordinate range test." + "reason": "Generation cache missing in chunked prefill vs flaky bbox-range test; unrelated areas." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Different Qwen-VL failures: from_pretrained weight_only error vs wrong vision_position_ids in video input handling." + "reason": "Qwen3VL weight_only load error vs Qwen2.5-VL video position-id mismatch; different model families and problems." }, { - "left": "issue:45081", - "right": "issue:45356", + "left": "issue:44857", + "right": "issue:47050", "accept": false, - "reason": "Both touch tokenizer regex handling, but one is a crash in _patch_mistral_regex and the other is a codec/warning regression in Kimi-K2.5." + "reason": "LwDetrImageLoss AMP crash vs CanineModel fp16 ONNX export dtype mismatch; separate model/export paths." }, { - "left": "issue:44857", - "right": "issue:47050", + "left": "issue:45081", + "right": "issue:45356", "accept": false, - "reason": "CUDA AMP loss crash is not the same as the fp16 ONNX export attention dtype mismatch." + "reason": "Mistral regex patch crash vs Kimi-K2.5 codec/regex regression; related tokenizer code, but distinct concrete failures." }, { "left": "issue:46032", "right": "issue:46463", "accept": false, - "reason": "Cache correctness with seq_len>1 is unrelated to T5 decoder-input error-message validation." + "reason": "Mamba2 cache/use_cache incorrect results vs T5 decoder-input validation message; different subsystems and behaviors." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "A tied-embeddings warning during LoRA fine-tuning is unrelated to incorrect visual encoder keys being saved." + "reason": "Tie-word-embeddings warning in fine-tuning vs Qwen3.5 save_pretrained visual-key regression; not the same issue." }, { - "left": "issue:43720", - "right": "issue:43819", + "left": "issue:39692", + "right": "issue:47050", "accept": false, - "reason": "Packed-weight loading failure is unrelated to DAC.from_latents missing STE behavior." + "reason": "SigLIP2 doc example bug vs CanineModel ONNX dtype mismatch; unrelated models and code paths." }, { - "left": "issue:43540", - "right": "issue:46294", + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Post-processor update at runtime vs LayoutLMv2 NER/padding crash; different tokenizer bugs." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 example/model mismatch vs quantization offloading behavior; no shared underlying defect." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "SigLIP2 AutoModel/pipeline issue vs CamemBERT masked LM regression; different models and failure modes." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regression warning vs list_repo_files revision forwarding in CI; unrelated functionality." + }, + { + "left": "issue:43720", + "right": "issue:43819", "accept": false, - "reason": "Different multimodal video failures with different models and failure modes; not one bug." + "reason": "BitNet packed-weights loading bug vs DAC missing-STE mismatch; different models and code paths." }, { "left": "issue:43596", "right": "issue:44220", "accept": false, - "reason": "Deepspeed/Bert index error and fbank feature extraction issue are unrelated." + "reason": "Deepspeed Zero-3 BertModel init error vs audio fbank feature extraction issue; unrelated." + }, + { + "left": "issue:43540", + "right": "issue:46294", + "accept": false, + "reason": "Qwen3OmniMoe video processing ValueError vs Qwen3.5 video chat-template token mismatch; different model/input handling bugs." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "torch.compile+SDPA failure is unrelated to DAC latent conversion mismatch." + "reason": "Bamba torch.compile SDPA failure vs DAC latent/forward mismatch; unrelated models and defects." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "Bamba compile/SDPA regression and DiaConfig save/load token ID loss are unrelated." + "reason": "Bamba compile/SDPA bug vs DiaConfig save/load token-id loss; not the same underlying problem." }, { "left": "issue:43295", "right": "issue:45834", "accept": false, - "reason": "processor.tokenizer regression and Kosmos2.5 long-OCR index error are different bugs." + "reason": "Processor/tokenizer regression vs Kosmos2.5 long OCR index error; different code paths and model families." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "Loading AutoImageProcessor from URL is unrelated to the non-template-node compile error." + "reason": "AutoImageProcessor URL loading vs template compilation TypeError; unrelated failure modes." }, { "left": "issue:45072", "right": "issue:47050", "accept": false, - "reason": "Both are dtype mismatches, but one is bfloat16 inference and the other is fp16 ONNX export; different paths." + "reason": "SwitchTransformers/TimmWrapper dtype mismatch vs CanineModel attention dtype mismatch; same broad symptom, but different models and fixes." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "ImageProcessor segmentation-map handling vs CustomTrainer compute_loss signature mismatch; unrelated components." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "BigBird special-token decode bug is unrelated to BitNet accelerate loading of packed weights." + "reason": "Tokenizer special-token registration bug vs BitNet packed-weight loading bug; different modules and fixes." }, { - "left": "issue:45356", - "right": "issue:45907", + "left": "issue:44062", + "right": "issue:44488", "accept": false, - "reason": "Tokenizer codec/regression in Kimi-K2.5 is unrelated to list_repo_files missing revision forwarding." + "reason": "AddedToken special-kwarg collision vs sleng-bert load failure; not the same bug." }, { - "left": "issue:39692", - "right": "issue:47050", + "left": "issue:44488", + "right": "issue:47020", "accept": false, - "reason": "SigLIP2 docs/example issues do not match the CanineModel fp16 ONNX attention dtype mismatch." + "reason": "Model load failure for sleng-bert vs XLMRobertaTokenizer/Unigram constructor type mismatch; different tokenizer/model issues." }, { - "left": "issue:39692", - "right": "issue:43873", + "left": "issue:44220", + "right": "issue:44661", "accept": false, - "reason": "SigLIP2 docs/quantization example errors are unrelated to the offloading regression." + "reason": "Audio fbank extraction issue vs add-new-model-like tokenizer-mapping failure; unrelated." }, { - "left": "issue:43994", - "right": "issue:44671", + "left": "issue:43531", + "right": "issue:43643", "accept": false, - "reason": "Different models and different symptoms: SigLIP2 nonsensical outputs vs CamemBERT masked LM prediction regression." + "reason": "Qwen3-MoE sliding_window bug vs trust_remote_code AutoConfig missing fields; different code paths and symptoms." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "Rejecting because the analysis overstates certainty on a few soft edges. In particular, issue:44514 and issue:44545 are clearly very similar, but the packet only supports a strong similarity/likely-duplicate judgment from titles, not an \"exact duplicate\" claim. Likewise, issue:42886 and issue:44843 look related, but the packet does not conclusively establish the same underlying bug/root cause from the issue titles alone. The rest of the pairwise rejections look conservative, but these stronger claims make the summary not fully grounded." - }, - "evaluator_used": true, - "retried": true, - "accepted_nontrivial_soft_edge": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -6152,31 +6151,28 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ - "issue:30064", "issue:34634", "issue:36010", - "issue:36331", "issue:39692", "issue:41720", + "issue:43322", "issue:43335", "issue:43526", - "issue:43531", "issue:43540", + "issue:43606", "issue:43611", + "issue:43618", "issue:43643", "issue:43653", + "issue:43697", "issue:43756", "issue:43819", "issue:43881", "issue:44008", - "issue:44062", "issue:44186", - "issue:44220", "issue:44442", "issue:44448", - "issue:44488", "issue:44561", - "issue:44661", "issue:44671", "issue:44805", "issue:44843", @@ -6185,11 +6181,15 @@ "issue:44936", "issue:44991", "issue:45020", + "issue:45072", "issue:45081", "issue:45381", + "issue:45685", "issue:45834", "issue:45987", "issue:46291", + "issue:46294", + "issue:46463", "issue:46650", "issue:46665", "issue:46693", @@ -6201,231 +6201,224 @@ "issue:47405" ], "soft_pairs": [ - "issue:30064|issue:36331", - "issue:44488|issue:47020", - "issue:44062|issue:44488", - "issue:43531|issue:43643", - "issue:44220|issue:44661", - "issue:45381|issue:46693", "issue:41720|issue:44991", + "issue:45381|issue:46693", "issue:43540|issue:44008", "issue:43611|issue:46665", - "issue:43526|issue:43881", "issue:43653|issue:43819", - "issue:46829|issue:46832", "issue:44186|issue:44898", + "issue:46829|issue:46832", "issue:34634|issue:36010", "issue:46650|issue:47405", "issue:44991|issue:47020", + "issue:43526|issue:43881", "issue:44008|issue:46858", - "issue:43643|issue:44877", "issue:44843|issue:45081", - "issue:45987|issue:46829", - "issue:39692|issue:45834", "issue:44561|issue:45020", + "issue:39692|issue:45834", + "issue:45987|issue:46829", + "issue:43643|issue:44877", "issue:43335|issue:43756", "issue:44186|issue:44442", "issue:44448|issue:44671", "issue:45987|issue:46932", + "issue:44991|issue:46291", + "issue:46294|issue:46463", "issue:44805|issue:44936", - "issue:44991|issue:46291" + "issue:43322|issue:45685", + "issue:43618|issue:43697", + "issue:43606|issue:45072" ], - "prepared_review_unit_hash": "fcbc44efb46985a97e7df5a5c6e9771fee28db2237f000517cfae9d7909f58a9", + "prepared_review_unit_hash": "f6cad95fb364dc7bdce5a08a6c4e4e7280ba971a65b284ab457321b5ff12761f", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 28, - "serialized_chars": 35877, - "estimated_input_tokens": 8970, - "estimated_eval_tokens": 18196 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36226, + "estimated_input_tokens": 9057, + "estimated_eval_tokens": 18370 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is not a duplicate cluster: the items span many unrelated bug reports (tokenizer loading, model config, trainer APIs, vision/audio preprocessing, cache behavior, offline mode, etc.). None of the soft pairs look like the same underlying bug or change.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "These issues span many unrelated models, tokenizers, configs, and backend/device paths. None of the proposed soft pairs look like true duplicates or should be merged into one PR/issue.", + "confidence": 0.99, + "canonical_issue_reason": "No single canonical issue fits this cluster: the items are heterogeneous and do not share one underlying bug or code path.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "No clear global best issue. The closest broad umbrella is 45020 for remote_code regressions, but it does not represent the many unrelated failures here.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:30064", - "right": "issue:36331", - "accept": false, - "reason": "Different subsystems and failure modes: image processor segmentation maps vs trainer compute_loss signature." - }, - { - "left": "issue:44488", - "right": "issue:47020", - "accept": false, - "reason": "Both are tokenizer-related, but they describe different models and different root causes (model loading failure vs Unigram constructor type mismatch)." - }, - { - "left": "issue:44062", - "right": "issue:44488", - "accept": false, - "reason": "One is an AddedToken special-arg bug; the other is a model-specific tokenizer loading failure." - }, - { - "left": "issue:43531", - "right": "issue:43643", - "accept": false, - "reason": "Qwen3-MoE sliding-window behavior is unrelated to AutoConfig trust_remote_code missing fields." - }, - { - "left": "issue:44220", - "right": "issue:44661", + "left": "issue:41720", + "right": "issue:44991", "accept": false, - "reason": "Audio fbank extraction bug vs tokenizer mapping/add-new-model failure; no shared code path." + "reason": "Different failures: Qwen3 auto device mapping CUDA assert vs tokenizer loading for EMBEDDIA/est-roberta." }, { "left": "issue:45381", "right": "issue:46693", "accept": false, - "reason": "Qwen2.5-VL video vision_position_ids bug is unrelated to FA2 long-context performance regression." - }, - { - "left": "issue:41720", - "right": "issue:44991", - "accept": false, - "reason": "CUDA assert during Qwen3 auto device mapping is not the same as tokenizer loading failure for est-roberta." + "reason": "Different code paths: Qwen2.5-VL video position IDs vs FA2 long-context performance regression." }, { "left": "issue:43540", "right": "issue:44008", "accept": false, - "reason": "Video input processing in Qwen3OmniMoe is unrelated to Gemma3n variable-name collision in forward()." + "reason": "Different models and bugs: Qwen3OmniMoe video ValueError vs Gemma3n forward variable collision." }, { "left": "issue:43611", "right": "issue:46665", "accept": false, - "reason": "Base_model_prefix loading breakage in v5 is distinct from inconsistent from_pretrained behavior inside Python classes." - }, - { - "left": "issue:43526", - "right": "issue:43881", - "accept": false, - "reason": "BeitImageProcessorFast label reduction bug is unrelated to glm-4v-9b loading failure." + "reason": "Both involve loading APIs, but the root causes differ: base_model_prefix breakage vs inconsistent from_pretrained inside classes." }, { "left": "issue:43653", "right": "issue:43819", "accept": false, - "reason": "BigBirdTokenizer special-token registration issue does not match DAC.from_latents vs forward STE mismatch." + "reason": "Unrelated subsystems: BigBirdTokenizer special-token decoding vs DAC.from_latents mismatch." }, { - "left": "issue:46829", - "right": "issue:46832", + "left": "issue:44186", + "right": "issue:44898", "accept": false, - "reason": "VideoPrism hidden_states output wiring and RTDetr feature-level mismatch are different model bugs." + "reason": "Different models and symptoms: LayoutLMv2Tokenizer NER/padding crash vs Perceiver image interpolation failure." }, { - "left": "issue:44186", - "right": "issue:44898", + "left": "issue:46829", + "right": "issue:46832", "accept": false, - "reason": "LayoutLMv2Tokenizer NER/padding crash is unrelated to Perceiver image classification with non-default resolution." + "reason": "Different RT-DETR/VideoPrism bugs: output field mismatch vs feature-level type error." }, { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "BarkProcessor voice_preset issue and GenerationMixin import error are separate model/library compatibility problems." + "reason": "Processor preset handling and a GenerationMixin import regression are unrelated bugs." }, { "left": "issue:46650", "right": "issue:47405", "accept": false, - "reason": "save_pretrained key rename corruption is unrelated to silent model_type mismatch on Auto class loading." + "reason": "Both touch model loading/saving, but one is reverse-rename corruption and the other is silent wrong-auto-class conversion skipping." }, { "left": "issue:44991", "right": "issue:47020", "accept": false, - "reason": "Different tokenizer failures on different models; no evidence of the same underlying bug." + "reason": "Different tokenizer implementations and failure modes; no shared underlying bug." }, { - "left": "issue:44008", - "right": "issue:46858", + "left": "issue:43526", + "right": "issue:43881", "accept": false, - "reason": "Gemma3n forward variable collision is unrelated to static cache seq-length/device tensor behavior." + "reason": "BeitImageProcessorFast label reduction bug vs glm-4v-9b loading failure are unrelated." }, { - "left": "issue:43643", - "right": "issue:44877", + "left": "issue:44008", + "right": "issue:46858", "accept": false, - "reason": "trust_remote_code AutoConfig missing fields is not the same as strict config blocking granite_speech loading." + "reason": "Gemma3n attribute collision and static-cache seq-length tensor arithmetic are separate code paths." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Offline model_info call in mistral regex patch and backend_tokenizer AttributeError are different failures in the same area but not the same bug." + "reason": "Same helper name, different bug: offline model_info call vs backend_tokenizer AttributeError." }, { - "left": "issue:45987", - "right": "issue:46829", + "left": "issue:44561", + "right": "issue:45020", "accept": false, - "reason": "StaticCache.get_seq_length() contract mismatch is unrelated to VideoPrism hidden_states output." + "reason": "Both mention trust_remote_code, but they are different regressions with different causes." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "SigLIP2 doc example errors and Kosmos2.5 OCR index error are distinct issues." + "reason": "Docs-example issues for SigLIP2 vs a long-OCR runtime error in Kosmos2.5 are not the same bug." }, { - "left": "issue:44561", - "right": "issue:45020", + "left": "issue:45987", + "right": "issue:46829", + "accept": false, + "reason": "StaticCache shape contract issue vs VideoPrism output misassignment are unrelated." + }, + { + "left": "issue:43643", + "right": "issue:44877", "accept": false, - "reason": "Both mention trust_remote_code/model loading, but one is is_torch_fx_available removal and the other is broader remote_code breakage; not the same concrete bug." + "reason": "Both are config/loading problems, but missing fields from trust_remote_code and strict granite_speech config are different issues." }, { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "SwitchTransformers sparse-layer creation bug is unrelated to Smollm3 RoPE layer dropping." + "reason": "Different architectures and failure modes: sparse encoder layer creation vs RoPE layer dropping." }, { "left": "issue:44186", "right": "issue:44442", "accept": false, - "reason": "LayoutLMv2Tokenizer NER crash and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are different tokenizer paths." + "reason": "Tokenizer crashes, but the affected tokenizers and root causes differ." }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Pegasus v4/v5 output differences and CamemBERT masked LM predictions are unrelated model behavior regressions." + "reason": "Model-specific v4/v5 behavior regressions, but not the same model or code path." }, { "left": "issue:45987", "right": "issue:46932", "accept": false, - "reason": "Static cache seq-length contract bug is unrelated to Gemma4 quantized embedder dtype casting." + "reason": "Static cache length contract vs Gemma4 quantized embedder dtype casting are unrelated." + }, + { + "left": "issue:44991", + "right": "issue:46291", + "accept": false, + "reason": "Tokenizer loading failure and kernels import-time dependency failure are separate problems." + }, + { + "left": "issue:46294", + "right": "issue:46463", + "accept": false, + "reason": "Qwen video chat-template mismatch and T5 decoder validation messaging are unrelated." }, { "left": "issue:44805", "right": "issue:44936", "accept": false, - "reason": "Mask shape/indexing error is not the same as trainer.evaluate() failing after trainer.train()." + "reason": "A tensor mask shape mismatch and trainer.evaluate() failing after train are not the same underlying issue." }, { - "left": "issue:44991", - "right": "issue:46291", + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Different backends/models: Llava Next segmentation fault vs MPS histogram support error." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "CLIPOutput attentions regression and RTDetrV2 output differences are unrelated model-specific bugs." + }, + { + "left": "issue:43606", + "right": "issue:45072", "accept": false, - "reason": "Tokenizer loading for est-roberta and pipelines import failing due to kernels version/revision requirements are different problems." + "reason": "CPU offload device mismatch and bfloat16 dtype mismatch are different failure modes in different models." } ] }, @@ -6453,17 +6446,16 @@ "issue:29159", "issue:34634", "issue:36331", - "issue:43322", + "issue:42175", + "issue:42915", + "issue:43388", "issue:43421", "issue:43540", "issue:43550", "issue:43577", "issue:43582", - "issue:43606", - "issue:43618", "issue:43644", "issue:43645", - "issue:43697", "issue:43716", "issue:43720", "issue:43742", @@ -6481,251 +6473,243 @@ "issue:44977", "issue:45020", "issue:45072", + "issue:45081", "issue:45183", "issue:45381", "issue:45479", "issue:45563", "issue:45584", - "issue:45685", "issue:45834", - "issue:46294", - "issue:46463", + "issue:46032", "issue:46512", "issue:46665", "issue:46752", "issue:46846", + "issue:47254", "issue:47276", + "issue:47333", "issue:47337" ], "soft_pairs": [ - "issue:43322|issue:45685", - "issue:46294|issue:46463", - "issue:43606|issue:45072", - "issue:43618|issue:43697", "issue:43716|issue:44008", - "issue:43550|issue:45072", "issue:45584|issue:45834", + "issue:43550|issue:45072", "issue:44479|issue:45834", - "issue:43782|issue:44977", - "issue:45183|issue:45479", "issue:43421|issue:44811", + "issue:45183|issue:45479", "issue:46752|issue:47337", + "issue:43782|issue:44977", "issue:43644|issue:43749", - "issue:43582|issue:46512", "issue:43577|issue:43818", + "issue:43582|issue:46512", "issue:45563|issue:47276", "issue:44610|issue:45072", - "issue:28218|issue:29159", - "issue:43720|issue:45072", "issue:34634|issue:36331", - "issue:43645|issue:46665", + "issue:43720|issue:45072", + "issue:28218|issue:29159", "issue:43540|issue:45381", + "issue:43645|issue:46665", "issue:43931|issue:46846", "issue:43749|issue:44291", "issue:47276|issue:47337", + "issue:43742|issue:45020", "issue:43825|issue:43994", - "issue:43742|issue:45020" - ], - "prepared_review_unit_hash": "13b8d9019b48d208c47dce6be1899d7ac3b760ae9f9c80816247b26411639a34", + "issue:42175|issue:45081", + "issue:47254|issue:47333", + "issue:43388|issue:46032", + "issue:42915|issue:47333" + ], + "prepared_review_unit_hash": "0db526fec7064d439f877e3a715ff000fab3acd41cd76cb1868f672ec855dd75", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 27, - "serialized_chars": 35380, - "estimated_input_tokens": 8845, - "estimated_eval_tokens": 17946 + "serialized_chars": 35531, + "estimated_input_tokens": 8883, + "estimated_eval_tokens": 18022 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are mostly unrelated issue reports across different models and subsystems. Only the tokenizer pair (28218, 29159) looks like a plausible duplicate/near-duplicate describing the same underlying added-token spacing inconsistency.", - "confidence": 0.84, - "canonical_issue_reason": "issue:29159 is the broader report: it explicitly frames the problem as inconsistent slow/fast tokenizer behavior, which likely subsumes the more specific extra-space-after-added-token symptom in issue:28218.", + "summary": "This cluster is a grab bag of largely unrelated transformer bugs (tokenizer, multimodal model loading, dtype/loading regressions, trainer/runtime issues). The soft similarities are mostly thematic, not the same underlying defect, so none of the proposed duplicate pairs should be merged.", + "confidence": 0.93, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:29159 is the best representative issue because it is the most general description of the tokenizer bug and covers the broader symptom space.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43322", - "right": "issue:45685", - "accept": false, - "reason": "Different failures: Llava Next loading segfault vs MPS histogram op not implemented for Int." - }, - { - "left": "issue:46294", - "right": "issue:46463", - "accept": false, - "reason": "Unrelated code paths: Qwen3.5 video chat-template mismatch vs T5 decoder input validation messaging." - }, - { - "left": "issue:43606", - "right": "issue:45072", - "accept": false, - "reason": "Both are runtime/device-dtype problems, but they affect different models and different bugs." - }, - { - "left": "issue:43618", - "right": "issue:43697", - "accept": false, - "reason": "CLIPOutput attentions regression is not the same as RTDetrV2 output divergence." - }, { "left": "issue:43716", "right": "issue:44008", "accept": false, - "reason": "Different model families and root causes: dtype mismatch vs variable-name collision in forward()." + "reason": "Both involve multimodal models, but one is a preprocessor/model dtype mismatch and the other is a Gemma3n forward variable-collision AttributeError; different code paths and fixes." }, { - "left": "issue:43550", - "right": "issue:45072", + "left": "issue:45584", + "right": "issue:45834", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is a different issue than dtype mismatch problems in other models." + "reason": "Whisper empty-transcription failure after special-token alignment is unrelated to Kosmos2.5 long-OCR index handling." }, { - "left": "issue:45584", - "right": "issue:45834", + "left": "issue:43550", + "right": "issue:45072", "accept": false, - "reason": "Empty transcription / align_special_tokens failure is unrelated to long OCR input indexing." + "reason": "One is a Bamba torch.compile/SDPA bug, the other is a dtype mismatch in different models; not the same concrete failure." }, { "left": "issue:44479", "right": "issue:45834", "accept": false, - "reason": "Qwen video-input regression and Kosmos2.5 OCR index error are different bugs." + "reason": "Qwen VL video-input regression and Kosmos2.5 long OCR index error are different model/input bugs." }, { - "left": "issue:43782", - "right": "issue:44977", + "left": "issue:43421", + "right": "issue:44811", "accept": false, - "reason": "Weight-only loading error for Qwen3VL is not the same as flash-attention generation failure in Qwen3.5." + "reason": "Both are tokenizer-adjacent, but runtime post-processor updates and Whisper batch_decode skip_special_tokens are separate behaviors." }, { "left": "issue:45183", "right": "issue:45479", "accept": false, - "reason": "Input validation XOR error and single-label classification zero-loss bug are different behaviors and code paths." + "reason": "Input validation XOR messaging is unrelated to the single-label-classification zero-loss bug." }, { - "left": "issue:43421", - "right": "issue:44811", + "left": "issue:46752", + "right": "issue:47337", "accept": false, - "reason": "Both involve tokenizer-adjacent behavior, but one is runtime post-processor updates and the other is batch_decode skip_special_tokens handling." + "reason": "Empty chat-template input crash and bbox-range flaky test are unrelated." }, { - "left": "issue:46752", - "right": "issue:47337", + "left": "issue:43782", + "right": "issue:44977", "accept": false, - "reason": "Empty apply_chat_template crash is unrelated to a flaky bbox-coordinate test." + "reason": "Qwen3VL weight_only loading error and Qwen3.5 flash-attention generation failure are different issues." }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Non-persistent buffer junk and FSDP CPU RAM efficient loading are different loading/runtime bugs." + "reason": "Both concern loading, but junk non-persistent buffers and broken FSDP CPU RAM efficient loading are distinct defects." }, { - "left": "issue:43582", - "right": "issue:46512", + "left": "issue:43577", + "right": "issue:43818", "accept": false, - "reason": "Apple Silicon TypeError in caching_allocator_warmup is unrelated to stale model.config.dtype after from_config(dtype=...)." + "reason": "BLIP2 dtype not propagating is unrelated to Video-LLaVA missing temporal attention/shared weights." }, { - "left": "issue:43577", - "right": "issue:43818", + "left": "issue:43582", + "right": "issue:46512", "accept": false, - "reason": "Blip2 dtype propagation bug is unrelated to Video-LLaVA temporal-attention/weight-sharing issues." + "reason": "Apple Silicon caching_allocator warmup TypeError and stale model.config.dtype after from_config are different dtype-related bugs." }, { "left": "issue:45563", "right": "issue:47276", "accept": false, - "reason": "Both are warning-message problems, but they concern different warnings in different generation modes." + "reason": "Both are warning-message problems, but about different APIs and different message text." }, { "left": "issue:44610", "right": "issue:45072", "accept": false, - "reason": "Processor image-size mismatch for OmDet-Turbo is not the same as dtype mismatch issues." + "reason": "OmDet-Turbo processor size mismatch is unrelated to SwitchTransformers/TimmWrapperModel bfloat16 dtype mismatches." }, { - "left": "issue:28218", - "right": "issue:29159", - "accept": true, - "reason": "Both describe the same tokenizer-added-token spacing inconsistency; 29159 is the broader slow-vs-fast tokenizer report." + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "BarkProcessor voice_preset and CustomTrainer.compute_loss signature mismatch are unrelated." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "BitNet packed-weight unpacking/loading failure is unrelated to dtype mismatch bugs." + "reason": "BitNet packed-weight unpacking during loading and a separate dtype mismatch in inference are not the same bug." }, { - "left": "issue:34634", - "right": "issue:36331", + "left": "issue:28218", + "right": "issue:29159", "accept": false, - "reason": "BarkProcessor voice_preset failure and CustomTrainer compute_loss signature error are completely different issues." + "reason": "Tokenizer extra-space after added token is a specific bug; the other is a broader slow/fast tokenizer inconsistency report." }, { - "left": "issue:43645", - "right": "issue:46665", + "left": "issue:43540", + "right": "issue:45381", "accept": false, - "reason": "Jupyter custom-model initialization regression is different from inconsistent from_pretrained behavior inside Python classes." + "reason": "Qwen3OmniMoe video-input ValueError and generic input_ids/inputs_embeds validation messaging are unrelated." }, { - "left": "issue:43540", - "right": "issue:45381", + "left": "issue:43645", + "right": "issue:46665", "accept": false, - "reason": "Qwen3OmniMoe video-input ValueError and qwen2.5-vl vision_position_ids regression are different video-input bugs." + "reason": "Jupyter custom-model initialization breakage and from_pretrained behavior inside Python classes are different loading scenarios with different causes." }, { "left": "issue:43931", "right": "issue:46846", "accept": false, - "reason": "Qwen3-VL weight shape mismatch and missing linear_attn TP plan entries are separate model-loading vs tensor-parallel bugs." + "reason": "Different Qwen variants and different failure modes: weight shape mismatch versus missing tp-plan entries causing OOM/shape errors." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "FSDP CPU RAM efficient loading and init_empty_weights unexpected argument errors are distinct loading-path issues." + "reason": "FSDP CPU RAM efficient loading and init_empty_weights unexpected argument are separate loading regressions." }, { "left": "issue:47276", "right": "issue:47337", "accept": false, - "reason": "A stale warning in continuous batching is unrelated to a flaky bbox coordinate range failure." + "reason": "Misleading synced_gpus warning and bbox coordinate range flaky test are unrelated." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "A key error loading one model is not the same underlying problem as broad remote_code breakage across recent versions." }, { "left": "issue:43825", "right": "issue:43994", "accept": false, - "reason": "Pipeline translation-support warning and siglip2 nonsensical outputs are different problems." + "reason": "Wrong translation-support warning text and SigLIP2 nonsensical outputs are unrelated." }, { - "left": "issue:43742", - "right": "issue:45020", + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Package dependency omission and a tokenizer backend_tokenizer AttributeError are unrelated." + }, + { + "left": "issue:47254", + "right": "issue:47333", "accept": false, - "reason": "MobileLLM key error on load and remote_code breakage are not the same underlying bug." + "reason": "PEFT+ZeRO-3 checkpointing error and disk-offloaded Qwen3 save failure are distinct training/offloading bugs." + }, + { + "left": "issue:42915", + "right": "issue:47333", + "accept": false, + "reason": "FineGrainedFP8Config failure for Qwen3Moe and disk-offload save failure for Qwen3-30B-A3B are unrelated." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The output is grounded in the packet and the soft-edge verdicts stay conservative. The only accepted pair, issue:28218 \u2194 issue:29159, is a plausible duplicate/near-duplicate, with 29159 reasonably serving as the broader tokenizer-behavior report. No soft-edge verdict appears overstated." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -6749,7 +6733,6 @@ "issue:43257", "issue:43262", "issue:43329", - "issue:43388", "issue:43452", "issue:43531", "issue:43550", @@ -6763,8 +6746,11 @@ "issue:43825", "issue:43827", "issue:43927", + "issue:44060", + "issue:44368", "issue:44589", "issue:44671", + "issue:44821", "issue:44857", "issue:44912", "issue:44936", @@ -6776,7 +6762,6 @@ "issue:45372", "issue:45406", "issue:45701", - "issue:46032", "issue:46175", "issue:46285", "issue:46291", @@ -6788,242 +6773,227 @@ "issue:47217", "issue:47254", "issue:47332", - "issue:47333", "issue:47337" ], "soft_pairs": [ - "issue:42175|issue:45081", - "issue:47254|issue:47333", - "issue:43388|issue:46032", "issue:46697|issue:46728", - "issue:42915|issue:47333", "issue:44671|issue:45701", - "issue:43577|issue:43582", "issue:44589|issue:45042", - "issue:43782|issue:45325", + "issue:43577|issue:43582", "issue:47254|issue:47332", + "issue:43782|issue:45325", "issue:43262|issue:47217", "issue:36010|issue:43645", "issue:46851|issue:47337", "issue:43262|issue:43819", - "issue:43582|issue:44857", "issue:43653|issue:44912", + "issue:43582|issue:44857", "issue:36010|issue:36331", - "issue:46175|issue:46932", "issue:43531|issue:45042", "issue:45003|issue:46285", + "issue:46175|issue:46932", "issue:43550|issue:44912", - "issue:45372|issue:46326", "issue:42175|issue:46291", + "issue:45372|issue:46326", "issue:43638|issue:44936", "issue:43819|issue:43927", "issue:43825|issue:43827", "issue:43257|issue:43329", "issue:43452|issue:44991", - "issue:45081|issue:45406" + "issue:42915|issue:44821", + "issue:45081|issue:45406", + "issue:44060|issue:44368" ], - "prepared_review_unit_hash": "b3ce4b2ad688b7821a8077d224c67fe82f8be97fcf6df2f50961ba462b1d546a", + "prepared_review_unit_hash": "71758b04dfe1d4a5b6a18af370e77563fcb180863923ae9fd0d720af70870a93", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36885, - "estimated_input_tokens": 9222, - "estimated_eval_tokens": 18700 + "soft_pair_count": 27, + "serialized_chars": 36290, + "estimated_input_tokens": 9073, + "estimated_eval_tokens": 18402 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are a broad mix of unrelated regressions and feature bugs across training, tokenizers, multimodal processing, quantization, offloading, and docs. None of the soft candidate pairs look like the same underlying bug/change, so no duplicates should be merged.", + "summary": "This cluster is mostly a bag of unrelated closed issues that only share broad lexical themes (v5 regressions, Qwen/Gemma/Gemma4, tokenizer/model loading, multimodal bugs). None of the soft-edge pairs look like the same underlying defect, so they should all stay separate.", "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue: the reports span unrelated bugs in different subsystems, so there is no meaningful duplicate anchor for the whole set.", + "canonical_issue_reason": "If a single issue must anchor the cluster, issue:45701 is the broadest/highest-signal user-facing regression report here (multiple references, general tokenization breakage). But it is not a true duplicate of the others.", "canonical_pr_reason": null, - "best_issue_reason": "No clear global best issue for deduplication; the items are mostly one-off, unrelated regressions rather than variants of one bug.", + "best_issue_reason": "issue:45701 is the most representative general regression report in this set; the rest are narrower, model-specific, or subsystem-specific bugs with no strong duplicate chain.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42175", - "right": "issue:45081", - "accept": false, - "reason": "Different problems: backend dependency/install packaging vs a tokenizer regex/attribute error." - }, - { - "left": "issue:47254", - "right": "issue:47333", - "accept": false, - "reason": "Different code paths: PEFT/ZeRO-3 gradient checkpointing failure vs disk-offload save failure for Qwen3-30B." - }, - { - "left": "issue:43388", - "right": "issue:46032", - "accept": false, - "reason": "Different subsystems: metric gathering label trimming vs Mamba2 cache correctness." - }, { "left": "issue:46697", "right": "issue:46728", "accept": false, - "reason": "Both are image-related but not the same bug: over-padding vs wrong patch-count calculation." + "reason": "Both are image-processor bugs, but one is over-padding in Swin2SR and the other is a wrong patch-count calculation in Aria; different code paths and symptoms." }, { - "left": "issue:42915", - "right": "issue:47333", + "left": "issue:44671", + "right": "issue:45701", "accept": false, - "reason": "Different Qwen3 issues: FP8 config failure vs saving a disk-offloaded model." + "reason": "One reports incorrect CamemBERT masked-LM predictions, the other a version-dependent tokenizer change; related theme, not the same bug." }, { - "left": "issue:44671", - "right": "issue:45701", + "left": "issue:44589", + "right": "issue:45042", "accept": false, - "reason": "Model prediction regression vs tokenizer-version/tokenization-change report; not the same bug." + "reason": "Float8 storage lookup failure and PIL backend requiring torchvision are unrelated runtime failures." }, { "left": "issue:43577", "right": "issue:43582", "accept": false, - "reason": "Unrelated failures: dtype loading for BLIP2 vs an Apple Silicon allocator TypeError." + "reason": "BLIP2 dtype staying float32 and Apple Silicon warmup TypeError are different models/platforms and different failure modes." }, { - "left": "issue:44589", - "right": "issue:45042", + "left": "issue:47254", + "right": "issue:47332", "accept": false, - "reason": "Different errors and layers: missing float8 storage vs PIL image processor torchvision dependency." + "reason": "PEFT+ZeRO-3 checkpointing error and offloading-save failure for Qwen3-VL are separate training vs serialization problems." }, { "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Same model family but different concrete bugs: weight_only loading error vs rope/index scaling logic." - }, - { - "left": "issue:47254", - "right": "issue:47332", - "accept": false, - "reason": "Different issues: training/checkpointing error vs inability to save a disk-offloaded model." + "reason": "Qwen3VL weight_only loading error and Qwen2.5-VL rope-index scaling bug are distinct model/version issues." }, { "left": "issue:43262", "right": "issue:47217", "accept": false, - "reason": "Different modalities and behaviors: audio chat-template sampling rate vs VL image-placeholder handling." + "reason": "Audio chat-template sampling-rate defaulting and VL image-placeholder handling are different modalities and different code paths." }, { "left": "issue:36010", "right": "issue:43645", "accept": false, - "reason": "Both involve imports/loading, but one is a GenerationMixin import error and the other is a custom-model notebook initialization regression." + "reason": "GenerationMixin import failure and custom-model initialization in notebooks are unrelated import/environment regressions." }, { "left": "issue:46851", "right": "issue:47337", "accept": false, - "reason": "Variable-length position_ids bug is unrelated to the flaky bbox coordinate range failure." + "reason": "Variable-length position_ids in Qwen3.5 and bbox range flakiness are unrelated bugs." }, { "left": "issue:43262", "right": "issue:43819", "accept": false, - "reason": "Unrelated audio template defaulting bug vs DAC latent/forward mismatch." - }, - { - "left": "issue:43582", - "right": "issue:44857", - "accept": false, - "reason": "Different platforms and failures: Apple Silicon warmup TypeError vs CUDA AMP loss crash." + "reason": "Audio processor chat-template sampling rate and DAC latent reconstruction mismatch are unrelated issues." }, { "left": "issue:43653", "right": "issue:44912", "accept": false, - "reason": "Tokenizer special-token decode bug vs MXFP4 quantization fallback issue." + "reason": "BigBirdTokenizer special-token registration and MXFP4 quantization fallback are unrelated tokenizer vs quantization bugs." }, { - "left": "issue:36010", - "right": "issue:36331", + "left": "issue:43582", + "right": "issue:44857", "accept": false, - "reason": "Both are API breakages, but one is an import failure and the other is a trainer compute_loss signature change." + "reason": "Apple Silicon allocator warmup TypeError and float16 AMP crash in LwDetrImageLoss are different platform/model failures." }, { - "left": "issue:46175", - "right": "issue:46932", + "left": "issue:36010", + "right": "issue:36331", "accept": false, - "reason": "Same model family, but one is a huge one-hot materialization problem and the other is dtype casting for quantized embedders." + "reason": "Importing GenerationMixin and CustomTrainer.compute_loss signature mismatch are unrelated API breakages." }, { "left": "issue:43531", "right": "issue:45042", "accept": false, - "reason": "Different bugs in different areas: Qwen3-MoE sliding window vs PIL backend torchvision requirement." + "reason": "Qwen3-MoE sliding-window behavior and PIL backend torchvision dependency are unrelated." }, { "left": "issue:45003", "right": "issue:46285", "accept": false, - "reason": "Different code paths: unsafe sys.modules access in modeling_utils vs CLIP text-model structure inconsistency." + "reason": "Unsafe sys.modules access in modeling_utils and CLIP text-model structure inconsistency are different internal API issues." }, { - "left": "issue:43550", - "right": "issue:44912", + "left": "issue:46175", + "right": "issue:46932", "accept": false, - "reason": "Different issues: torch.compile/SDPA failure vs MXFP4 quantization loading fallback." + "reason": "The first is a memory blow-up in Gemma4 vision patch embeddings; the second is a dtype-casting bug for quantized embedders." }, { - "left": "issue:45372", - "right": "issue:46326", + "left": "issue:43550", + "right": "issue:44912", "accept": false, - "reason": "Both touch mistral_common, but one is an import-time processor loading breakage and the other is a continue_final_message behavior bug." + "reason": "Bamba torch.compile SDPA failure and MXFP4 quantization fallback are unrelated problems." }, { "left": "issue:42175", "right": "issue:46291", "accept": false, - "reason": "Different dependency/import failures: torch extra missing tensorflow vs kernels package version/revision requirement." + "reason": "Missing TensorFlow in torch-only install and kernels version/revision import failure are unrelated packaging issues." + }, + { + "left": "issue:45372", + "right": "issue:46326", + "accept": false, + "reason": "Both involve mistral_common-adjacent processor/tokenizer code, but one is an ImportError for ReasoningEffort and the other is continue_final_message behavior; not the same defect." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "Different training failures: zero3/non-pretrained Bert IndexError vs evaluate-after-train behavior." + "reason": "ZeRO-3 index error with a non-pretrained Bert model and trainer.evaluate() failing after train() are unrelated training-loop issues." }, { "left": "issue:43819", "right": "issue:43927", "accept": false, - "reason": "Different models and mechanisms: DAC latent mismatch vs DiaConfig losing custom token IDs after save/load." + "reason": "DAC latent reconstruction mismatch and DiaConfig losing custom token IDs are different components and different failure modes." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "Related theme, but not the same bug: runtime pipeline error message vs docs still referencing removed pipeline usage." + "reason": "One is an error-message regression about translation support in pipeline(), the other is stale docs still referencing pipeline(); related theme, not one bug." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "Different areas: Qwen3 MoE weight conversion during loading vs multimodal video token counting undefined variables." + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed and a multimodal token-counting undefined-name bug are unrelated." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "Different tokenizer-loading failures: gguf_file integration breakage vs a model-specific tokenizer regression in v5." + "reason": "gguf_file breakage for AutoTokenizer/AutoModel loading and a specific tokenizer-loading failure for EMBEDDIA/est-roberta are separate pretrained-loading problems." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and AutoImageProcessor-from-URL failure are unrelated." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Both are attribute errors around tokenizer/processor internals, but they affect different objects and code paths." + "reason": "Mistral regex patching on a tokenizer without backend_tokenizer and Gemma4Processor missing _tokenizer are distinct tokenizer/processor bugs." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention tied-weights warnings in Qwen-family models, but they describe different models and different causes; not the same concrete bug." } ] }, @@ -7060,6 +7030,7 @@ "issue:43329", "issue:43425", "issue:43450", + "issue:43525", "issue:43526", "issue:43643", "issue:43696", @@ -7069,9 +7040,7 @@ "issue:43854", "issue:43906", "issue:43931", - "issue:44060", "issue:44291", - "issue:44368", "issue:44387", "issue:44479", "issue:44554", @@ -7094,248 +7063,249 @@ "issue:46291", "issue:47020", "issue:47050", + "issue:47145", "issue:47333" ], "soft_pairs": [ - "issue:42915|issue:44821", - "issue:44060|issue:44368", "issue:43526|issue:43697", - "issue:43854|issue:44977", "issue:43257|issue:43696", - "issue:28370|issue:29159", + "issue:43854|issue:44977", "issue:45859|issue:47050", + "issue:28370|issue:29159", "issue:43746|issue:44811", + "issue:42915|issue:43742", "issue:45084|issue:45538", "issue:38175|issue:44610", "issue:43906|issue:46209", - "issue:42915|issue:43742", - "issue:43262|issue:44811", "issue:43329|issue:44479", + "issue:43262|issue:44811", "issue:44291|issue:45042", "issue:42490|issue:43257", "issue:43643|issue:45759", "issue:43329|issue:45600", - "issue:43526|issue:44554", - "issue:34567|issue:40990", - "issue:39692|issue:44912", "issue:43329|issue:43450", + "issue:39692|issue:44912", + "issue:34567|issue:40990", + "issue:43526|issue:44554", "issue:44387|issue:44617", "issue:44589|issue:47333", "issue:29159|issue:34567", - "issue:45081|issue:47020", "issue:43257|issue:43931", + "issue:45081|issue:47020", "issue:43425|issue:46291", "issue:44987|issue:45538", - "issue:44291|issue:44821" + "issue:44291|issue:44821", + "issue:43525|issue:46291", + "issue:44912|issue:47145" ], - "prepared_review_unit_hash": "a29df9c4346682e91d9c7f4fc680480a7012b760548a69b1e7d4c06748d81152", + "prepared_review_unit_hash": "76d3e91e479b3d6b3078f0bc6cc118c9e0b804292b87b75200bf32e723e5f58d", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36868, - "estimated_input_tokens": 9217, - "estimated_eval_tokens": 18690 + "serialized_chars": 36781, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is highly heterogeneous: the soft pairs share broad subsystem keywords, but they do not describe the same underlying bug or change. No PRs are present. I\u2019d use issue 43329 as the closest triage anchor, but it is not a true duplicate target for the rest of the set.", - "confidence": 0.94, - "canonical_issue_reason": "issue:43329 is the most concrete, code-path-specific bug here and the closest thing to a reusable anchor for the multimodal/video reports.", + "summary": "This cluster is a grab-bag of unrelated Transformers issues that only share broad themes like loading, tokenizers, multimodal/video, quantization, or OOMs. None of the soft pairs look like the same underlying bug or a safe PR merge candidate.", + "confidence": 0.95, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43329 is the best overall triage candidate because it pinpoints an exact failing branch/helper in a broadly impactful VLM/video path.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:42915", - "right": "issue:44821", - "accept": false, - "reason": "Different failures: a Qwen3MoE FP8 loading/config issue vs loading AutoImageProcessor from a URL." - }, - { - "left": "issue:44060", - "right": "issue:44368", + "left": "issue:43526", + "right": "issue:43697", "accept": false, - "reason": "Both mention tied-weights warnings, but they point to different models and different code paths; not the same bug." + "reason": "Different models and code paths: BeitImageProcessorFast label reduction vs RTDetrV2 output drift." }, { - "left": "issue:43526", - "right": "issue:43697", + "left": "issue:43257", + "right": "issue:43696", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug vs RTDetrV2 output drift; unrelated symptoms and fixes." + "reason": "One is Qwen3 MOE weight conversion with accelerate/deepspeed; the other is a GPT-oss CUDA OOM. Same broad area, not the same defect." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "One is a model-load failure in tests, the other is a flash-attention generation regression; not the same issue." + "reason": "Different models and symptoms: GLM-4.7-Flash loading in tests vs Qwen3.5 flash-attention generation problems." }, { - "left": "issue:43257", - "right": "issue:43696", + "left": "issue:45859", + "right": "issue:47050", "accept": false, - "reason": "MoE weight conversion/loading problem vs CUDA OOM; different root causes." + "reason": "Unrelated failures: CPU-offload compatibility in Qwen3.5Moe rotary embedding vs CanineModel ONNX dtype mismatch." }, { "left": "issue:28370", "right": "issue:29159", "accept": false, - "reason": "Tokenizer-related, but one is missing vocab_file for custom SentencePiece models and the other is slow/fast tokenizer inconsistency." + "reason": "Both tokenizer-related, but one is missing vocab_file for custom SentencePiece and the other is slow/fast tokenizer inconsistency." }, { - "left": "issue:45859", - "right": "issue:47050", + "left": "issue:43746", + "right": "issue:44811", "accept": false, - "reason": "CPU-offload rotary embedding incompatibility vs fp16 ONNX attention dtype mismatch; different code paths." + "reason": "Different functionality: GraniteSpeech local PEFT checkpoint loading vs Whisper batch_decode skip_special_tokens handling." }, { - "left": "issue:43746", - "right": "issue:44811", + "left": "issue:42915", + "right": "issue:43742", "accept": false, - "reason": "PEFT/local checkpoint loading failure vs batch_decode ignoring skip_special_tokens; unrelated behavior." + "reason": "Different model/config bugs: Qwen3Moe FineGrainedFP8Config failure vs MobileLLM-125M key error." }, { "left": "issue:45084", "right": "issue:45538", "accept": false, - "reason": "Template compilation error vs CLIPTokenizer max_length bug; no shared underlying defect." + "reason": "No shared root cause: template compilation error vs CLIPTokenizer model_max_length being too large." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Both involve vision model/processor mismatch, but one is zero probabilities on SigLIP2 and the other is an image-size mismatch in OmDet-Turbo." + "reason": "Different models and failure modes: SigLIP2 zero probabilities vs OmDet-Turbo processor/model image-size mismatch." }, { "left": "issue:43906", "right": "issue:46209", "accept": false, - "reason": "An isolated repro of an older issue is not the same as an FP8 inference failure on B200." + "reason": "Too generic to unify: an isolated reproduction of an earlier issue vs an FP8 inference failure on B200 hardware." }, { - "left": "issue:42915", - "right": "issue:43742", + "left": "issue:43329", + "right": "issue:44479", "accept": false, - "reason": "Different loading problems for different models; no shared failing code path." + "reason": "Both touch video/multimodal handling, but one is undefined symbols in _get_num_multimodal_tokens and the other is a broader v5.3.0 video regression across several models." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Audio chat-template sampling-rate default vs Whisper batch_decode token filtering; unrelated bugs." - }, - { - "left": "issue:43329", - "right": "issue:44479", - "accept": false, - "reason": "Both are video/multimodal regressions, but one is an undefined-name bug in _get_num_multimodal_tokens and the other is a broader v5.3.0 video-input regression." + "reason": "Different audio bugs: apply_chat_template default sampling rate vs Whisper decode ignoring skip_special_tokens." }, { "left": "issue:44291", "right": "issue:45042", "accept": false, - "reason": "init_empty_weights TypeError vs PIL backend incorrectly requiring torchvision; unrelated." + "reason": "Different subsystems: init_empty_weights argument handling vs PIL backend image processors wrongly requiring torchvision." }, { "left": "issue:42490", "right": "issue:43257", "accept": false, - "reason": "Local-path import warning with trailing slash vs Qwen3 MoE weight conversion failure; different problems." + "reason": "Local-path slash import warning and Qwen3 MOE weight-conversion failure are unrelated." }, { "left": "issue:43643", "right": "issue:45759", "accept": false, - "reason": "Both concern composite configs, but one is missing fields from trust_remote_code config loading and the other is failure to unwrap text_config for from_config." + "reason": "Both are config-loading issues, but the concrete bugs differ: missing fields under trust_remote_code vs not unwrapping text_config in composite multimodal configs." }, { "left": "issue:43329", "right": "issue:45600", "accept": false, - "reason": "Video token-counting bug vs stale auto_mappings reference to removed configs; unrelated." + "reason": "Video token counting bug vs stale auto_mappings references to removed Sam3LiteText configs; different files and causes." }, { - "left": "issue:43526", - "right": "issue:44554", + "left": "issue:43329", + "right": "issue:43450", "accept": false, - "reason": "Beit label reduction bug vs MPS attention correctness issue; different subsystems." + "reason": "Both mention video, but one is undefined helper usage in token counting and the other is batched video processor shape output." }, { - "left": "issue:34567", - "right": "issue:40990", + "left": "issue:39692", + "right": "issue:44912", "accept": false, - "reason": "Trainer token-count tracking vs high perplexity on GPT-oss-20b; not the same defect." + "reason": "Different problem domains: SigLIP2 documentation errors vs GPT-oss MXFP4 quantization fallback to bf16." }, { - "left": "issue:39692", - "right": "issue:44912", + "left": "issue:34567", + "right": "issue:40990", "accept": false, - "reason": "SigLIP2 doc-example errors vs MXFP4 quantization loading fallback; only loosely model-related." + "reason": "Trainer token accounting bug vs extreme perplexity on gpt-oss; no shared underlying code path." }, { - "left": "issue:43329", - "right": "issue:43450", + "left": "issue:43526", + "right": "issue:44554", "accept": false, - "reason": "Undefined helpers in video token counting vs batched video processor shape bug; distinct processor issues." + "reason": "Image processor label reduction bug and MPS attention correctness issue are unrelated." }, { "left": "issue:44387", "right": "issue:44617", "accept": false, - "reason": "Int4 reserved-memory regression vs Sam3Video OOM; same symptom class, different root causes." + "reason": "Both can surface as OOM, but one is int4 quantization memory reservation and the other is a Sam3Video runtime OOM." }, { "left": "issue:44589", "right": "issue:47333", "accept": false, - "reason": "Float8 storage lookup error vs inability to save disk-offloaded model; unrelated offloading/storage problems." + "reason": "Storage lookup failure for FP8 tensors vs saving a disk-offloaded Qwen3 model; not the same bug." }, { "left": "issue:29159", "right": "issue:34567", "accept": false, - "reason": "Tokenizer inconsistency vs trainer token counter not updating; no shared underlying bug." + "reason": "Tokenizer slow/fast inconsistency is unrelated to TrainerState num_input_tokens_seen not updating." }, { - "left": "issue:45081", - "right": "issue:47020", + "left": "issue:43257", + "right": "issue:43931", "accept": false, - "reason": "Mistral regex patch crash vs XLMRoberta Unigram init type mismatch; both tokenizer loading issues but different failures." + "reason": "Same model family name, but different failures: weight conversion with accelerate/deepspeed vs Qwen3-VL weight-shape mismatch on load." }, { - "left": "issue:43257", - "right": "issue:43931", + "left": "issue:45081", + "right": "issue:47020", "accept": false, - "reason": "Qwen3 MoE loading/convert issue vs Qwen3-VL weight-shape mismatch; different model families and causes." + "reason": "Both are tokenizer construction issues, but they involve different tokenizers and different failure mechanisms." }, { "left": "issue:43425", "right": "issue:46291", "accept": false, - "reason": "Torch version incompatibility vs kernels package import/version requirement; separate dependency problems." + "reason": "Unrelated dependency problems: Torch 2.10 incompatibility vs kernels v0.15 import requirement." }, { "left": "issue:44987", "right": "issue:45538", "accept": false, - "reason": "Model loading failure for a specific repo vs CLIPTokenizer model_max_length bug; not the same issue." + "reason": "Different components and bugs: loading physical-intelligence/fast on transformers>=5.1.0 vs CLIPTokenizer max-length misconfiguration." }, { "left": "issue:44291", "right": "issue:44821", "accept": false, - "reason": "init_empty_weights argument error vs loading AutoImageProcessor from URL; unrelated." + "reason": "init_empty_weights TypeError and AutoImageProcessor URL loading are separate issues." + }, + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Llama4Config missing pad_token_id is unrelated to kernels-v0.15 pipeline import failure." + }, + { + "left": "issue:44912", + "right": "issue:47145", + "accept": false, + "reason": "Both are quantization-related, but one is MXFP4 falling back to bf16 and the other is silent corruption from missing weight_scale on frozen FP8 checkpoints." } ] }, @@ -7371,7 +7341,6 @@ "issue:42915", "issue:43404", "issue:43454", - "issue:43525", "issue:43531", "issue:43540", "issue:43577", @@ -7382,14 +7351,15 @@ "issue:43749", "issue:43824", "issue:43906", + "issue:44062", "issue:44291", "issue:44479", "issue:44568", "issue:44589", "issue:44756", "issue:44821", - "issue:44912", "issue:44936", + "issue:44991", "issue:45127", "issue:45137", "issue:45356", @@ -7398,221 +7368,220 @@ "issue:45459", "issue:45538", "issue:45803", - "issue:46291", "issue:46421", "issue:46424", "issue:46693", "issue:46701", "issue:46821", - "issue:47145", + "issue:46869", "issue:47254", "issue:47332" ], "soft_pairs": [ - "issue:43525|issue:46291", - "issue:44912|issue:47145", "issue:45459|issue:45803", - "issue:43540|issue:46701", "issue:43742|issue:44291", + "issue:43540|issue:46701", "issue:28282|issue:34689", - "issue:43454|issue:43577", "issue:28282|issue:29870", - "issue:43749|issue:44756", + "issue:43454|issue:43577", "issue:33357|issue:34567", - "issue:43650|issue:43906", + "issue:43749|issue:44756", "issue:44589|issue:47332", + "issue:43650|issue:43906", "issue:42175|issue:43824", "issue:43531|issue:44821", "issue:44568|issue:45356", - "issue:46421|issue:46424", "issue:44479|issue:46693", - "issue:24643|issue:29159", + "issue:46421|issue:46424", "issue:42915|issue:45538", + "issue:24643|issue:29159", "issue:43582|issue:47254", "issue:44936|issue:46821", - "issue:43404|issue:45127", "issue:45137|issue:45405", + "issue:43404|issue:45127", "issue:28282|issue:29127", - "issue:43697|issue:45357" + "issue:43697|issue:45357", + "issue:44062|issue:44991", + "issue:45538|issue:46869" ], - "prepared_review_unit_hash": "68b4e1b2bae6180d6fd1a713efaa6bf8d9b1b2c34cd1c31809e56e1030b487d2", + "prepared_review_unit_hash": "88fec16a666bba52edee3f9688b8056827c27631145ee4ca9bd371603f1314ff", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 25, - "serialized_chars": 35267, - "estimated_input_tokens": 8817, - "estimated_eval_tokens": 17890 + "serialized_chars": 34614, + "estimated_input_tokens": 8654, + "estimated_eval_tokens": 17564 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated issues across different models, loaders, tokenizers, caching, and training paths. None of the soft pairs look like the same underlying bug, so I would not merge any of them as duplicates.", - "confidence": 0.98, - "canonical_issue_reason": null, + "summary": "This cluster is dominated by superficial text similarity rather than true duplicates. The paired issues mostly span unrelated subsystems or root causes: tokenizer errors, multimodal model loading, cache behavior, DeepSpeed/FSDP, and model-specific regressions. I would not merge any of the soft pairs.", + "confidence": 0.89, + "canonical_issue_reason": "No clear duplicate hub emerges from the cluster; the issues are too heterogeneous. If one must be picked as a representative, issue 28282 is the clearest standalone user-facing failure report, but it does not unify the rest of the set.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 28282 is the most concise and broadly understandable report, with an explicit error message and a clear reproduction context. It is representative as a high-signal issue, but not as a duplicate anchor for the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43525", - "right": "issue:46291", - "accept": false, - "reason": "Different failures: missing `pad_token_id` on Llama4Config vs a kernels dependency/version requirement during pipeline import." - }, - { - "left": "issue:44912", - "right": "issue:47145", + "left": "issue:45459", + "right": "issue:45803", "accept": false, - "reason": "Both are quantization-related, but one is MXFP4 fallback to bf16 and the other is FP8 frozen checkpoint loading without `weight_scale`; different code paths and symptoms." + "reason": "Both involve error handling, but they hit different code paths and different failure modes; not the same underlying bug." }, { - "left": "issue:45459", - "right": "issue:45803", + "left": "issue:43742", + "right": "issue:44291", "accept": false, - "reason": "Both mention broad exception handling, but they affect different functions and different failure modes; not the same bug." + "reason": "Both are model-loading failures, but one is a missing key and the other is an init_empty_weights argument mismatch; distinct causes." }, { "left": "issue:43540", "right": "issue:46701", "accept": false, - "reason": "Different model families and different pipeline/input types; no shared concrete code-path problem." + "reason": "Different multimodal pipelines and different failure surfaces; only broad subsystem overlap." }, { - "left": "issue:43742", - "right": "issue:44291", + "left": "issue:28282", + "right": "issue:34689", "accept": false, - "reason": "Both are loading errors, but one is a key error for MobileLLM and the other is an unexpected constructor argument with `init_empty_weights`." + "reason": "One is an environment/dependency import error, the other a model-loading regression for a specific vision model." }, { "left": "issue:28282", - "right": "issue:34689", + "right": "issue:29870", "accept": false, - "reason": "Generic loading/import issues, but one is missing PyTorch and the other is a specific Llama 3.2 Vision model-loading regression." + "reason": "Completely different bug classes: missing PyTorch vs decoder attention mask assignment in generation." }, { "left": "issue:43454", "right": "issue:43577", "accept": false, - "reason": "Both concern model loading correctness, but one is missing weight tying in AyaVision/Mistral3 and the other is dtype propagation in BLIP2." + "reason": "Both touch model internals, but one is weight tying for a specific architecture and the other is dtype propagation on BLIP2 loading." }, { - "left": "issue:28282", - "right": "issue:29870", + "left": "issue:33357", + "right": "issue:34567", "accept": false, - "reason": "Completely different problems: missing PyTorch vs a BART `generate()` unbound local error." + "reason": "Unrelated areas: a MacOS CLIP bus error versus a TrainerState bookkeeping bug." }, { "left": "issue:43749", "right": "issue:44756", "accept": false, - "reason": "Different infrastructure issues: FSDP CPU RAM efficient loading vs mmap-related OOM on Strix Halo." + "reason": "Both relate to memory/loading, but one is FSDP CPU RAM-efficient loading and the other is mmap/OOM on a specific machine family." }, { - "left": "issue:33357", - "right": "issue:34567", + "left": "issue:44589", + "right": "issue:47332", "accept": false, - "reason": "Different subsystems and symptoms: MacOS bus error in CLIP model loading vs TrainerState token count tracking." + "reason": "Different storage/offload failures; no shared code-path is evident." }, { "left": "issue:43650", "right": "issue:43906", "accept": false, - "reason": "No clear shared bug; one is a placeholder-style issue, the other is a reproduction of an unrelated external issue." - }, - { - "left": "issue:44589", - "right": "issue:47332", - "accept": false, - "reason": "Different storage/quantization error vs disk-offload save failure; not the same underlying defect." + "reason": "No meaningful technical overlap beyond both being issue reports." }, { "left": "issue:42175", "right": "issue:43824", "accept": false, - "reason": "Backend packaging/install issue vs a missing class import; unrelated root causes." + "reason": "Package/backend installation problem versus missing exported class import; not the same defect." }, { "left": "issue:43531", "right": "issue:44821", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE vs loading an image processor from URL; different code paths." + "reason": "Different model families and unrelated failures; one is sliding_window behavior, the other image processor loading from URL." }, { "left": "issue:44568", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different tokenizers/models and different logic." - }, - { - "left": "issue:46421", - "right": "issue:46424", - "accept": false, - "reason": "Both are cache-related, but one is lazy init causing recompilation and the other is max_cache_length handling for quantized cache only." + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." }, { "left": "issue:44479", "right": "issue:46693", "accept": false, - "reason": "Different regressions: video-input handling for several Qwen VL models vs FA2 long-context performance regression." + "reason": "One is a video-input regression across Qwen VL variants; the other is a performance regression with FA2 and long-context generation." }, { - "left": "issue:24643", - "right": "issue:29159", + "left": "issue:46421", + "right": "issue:46424", "accept": false, - "reason": "DeepSpeed training weight-dimension error vs tokenizer fast/slow inconsistency; unrelated." + "reason": "Both mention cache, but one is lazy-init causing recompilation and the other is max_cache_length being ignored; distinct bugs." }, { "left": "issue:42915", "right": "issue:45538", "accept": false, - "reason": "FineGrainedFP8Config failure in Qwen3Moe vs CLIPTokenizer max length constant; no shared bug." + "reason": "No real overlap: FP8 configuration failure versus CLIP tokenizer max_length behavior." + }, + { + "left": "issue:24643", + "right": "issue:29159", + "accept": false, + "reason": "Different domains entirely: DeepSpeed training weight-shape error versus tokenizer slow/fast inconsistency." }, { "left": "issue:43582", "right": "issue:47254", "accept": false, - "reason": "Different training/runtime failures: Apple Silicon allocator warmup TypeError vs PEFT+ZeRO-3 checkpointing error." + "reason": "Both touch distributed or Apple/PEFT contexts, but the actual failures and code paths are unrelated." }, { "left": "issue:44936", "right": "issue:46821", "accept": false, - "reason": "Trainer evaluate-after-train bug vs TimesFM `window_size` attribute error; unrelated." + "reason": "Trainer state/eval sequencing bug versus a TimesFM window_size attribute error; not duplicates." }, { - "left": "issue:43404", - "right": "issue:45127", + "left": "issue:45137", + "right": "issue:45405", "accept": false, - "reason": "Both involve tied weights, but one is missing `lm_head` tying in Mistral3 and the other is LoRA merge collapse after extended-vocab changes." + "reason": "DeepSpeed ZeRO-3 deque failure and an unreleased PEFT version bump are unrelated." }, { - "left": "issue:45137", - "right": "issue:45405", + "left": "issue:43404", + "right": "issue:45127", "accept": false, - "reason": "DeepSpeed ZeRO-3 deque error vs a PEFT version bump/release mismatch; not the same issue." + "reason": "Both involve tied embeddings/lm_head, but one is a missing weight-tie implementation and the other is a LoRA merge with extended vocabulary; different root causes." }, { "left": "issue:28282", "right": "issue:29127", "accept": false, - "reason": "Missing PyTorch import error vs LayoutLMv3 error-message clarity; unrelated." + "reason": "Missing PyTorch import versus LayoutLMv3 box-validation messaging; no duplicate evidence." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "Different model families and failures: RTDetrV2 output drift in v5 vs Qwen3.5 save_pretrained key regression." + "reason": "One is an inference-output regression, the other a save_pretrained key serialization bug; different behaviors and fixes." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Tokenizer-related failures, but one is an AddedToken argument conflict and the other is a model-specific tokenizer loading regression." + }, + { + "left": "issue:45538", + "right": "issue:46869", + "accept": false, + "reason": "CLIP tokenizer max_length bug and YOLO multi-GPU device-map failure are unrelated." } ] }, @@ -7669,263 +7638,249 @@ "issue:44898", "issue:44964", "issue:44987", - "issue:44991", "issue:45042", "issue:45081", "issue:45325", "issue:45381", "issue:45488", - "issue:45538", "issue:45701", + "issue:45823", "issue:45907", + "issue:46463", "issue:46512", "issue:46701", "issue:46832", - "issue:46869", "issue:47332", "issue:47333" ], "soft_pairs": [ - "issue:44062|issue:44991", - "issue:45538|issue:46869", "issue:44857|issue:46832", - "issue:39379|issue:44554", - "issue:44568|issue:45488", "issue:41720|issue:43540", "issue:43540|issue:44479", - "issue:42915|issue:43749", + "issue:39379|issue:44554", + "issue:44568|issue:45488", "issue:43540|issue:44560", "issue:42915|issue:47332", + "issue:45042|issue:45081", + "issue:42915|issue:43749", "issue:27957|issue:28282", "issue:43452|issue:43582", - "issue:45042|issue:45081", "issue:34567|issue:36331", "issue:44479|issue:45381", "issue:44062|issue:45081", - "issue:43577|issue:46512", - "issue:38175|issue:44898", "issue:43742|issue:44589", "issue:42175|issue:44488", + "issue:38175|issue:44898", "issue:43493|issue:43994", + "issue:43577|issue:46512", "issue:28282|issue:28370", "issue:44792|issue:44964", "issue:43742|issue:47333", - "issue:44987|issue:45907", "issue:44849|issue:45701", - "issue:43756|issue:45325", - "issue:43716|issue:46701" + "issue:44987|issue:45907", + "issue:43716|issue:46701", + "issue:45823|issue:46463", + "issue:43756|issue:45325" ], - "prepared_review_unit_hash": "122a086ea81369a1d37ffb926dde8ea6f4991410377afd320119e991656cceea", + "prepared_review_unit_hash": "19fd2a16fe5486438065d56d76c03eb5a3b5d4cf7241364338deb43a7bb24c90", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36043, - "estimated_input_tokens": 9011, - "estimated_eval_tokens": 18278 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35219, + "estimated_input_tokens": 8805, + "estimated_eval_tokens": 17866 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous and does not form a broad duplicate set. The only likely duplicate pair is the SigLIP2 reports (43493/43994), which describe the same underlying model correctness problem from different angles.", - "confidence": 0.86, - "canonical_issue_reason": "No single cluster-wide canonical issue: most items are unrelated. If one representative is needed, 43994 is the strongest candidate because it captures the user-facing symptom of the only likely duplicate pair.", + "summary": "The cluster is mostly a set of unrelated model-specific bugs, with a clear Qwen video-input regression subgroup centered on issue:44479. A SigLIP2-related subgroup also appears, but there is not enough evidence to promote any PR canonical here.", + "confidence": 0.81, + "canonical_issue_reason": "issue:44479 is the broadest and most central report in the clearest duplicate family, covering the shared Qwen video-input regression across several related model variants.", "canonical_pr_reason": null, - "best_issue_reason": "43994 is the best standalone issue to represent the cluster\u2019s only plausible duplicate pair, since it states the concrete bad-output symptom for SigLIP2.", + "best_issue_reason": "issue:44479 best represents the cluster because it anchors the most coherent duplicate subgroup while the rest of the issues are largely fragmented and unrelated.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44062", - "right": "issue:44991", + "left": "issue:44857", + "right": "issue:46832", "accept": false, - "reason": "Different tokenizer bugs: AddedToken argument collision vs model-specific tokenizer loading failure." + "reason": "Different detector models and different failure modes; both are vision bugs but not the same underlying issue." }, { - "left": "issue:45538", - "right": "issue:46869", + "left": "issue:41720", + "right": "issue:43540", "accept": false, - "reason": "CLIPTokenizer max-length default is unrelated to YOLO multi-GPU device-map failure." + "reason": "Both mention Qwen3, but one is an auto device-mapping CUDA assert and the other is a video-input processing error." }, { - "left": "issue:44857", - "right": "issue:46832", + "left": "issue:43540", + "right": "issue:44479", "accept": false, - "reason": "Different detection-model problems: AMP crash in LwDetrImageLoss vs RT-DETR feature-level mismatch." + "reason": "Related video-input failures in nearby Qwen model lines, but the packet does not show enough evidence that they are the same regression family." }, { "left": "issue:39379", "right": "issue:44554", "accept": false, - "reason": "mllama mask handling and MPS attention/value-head dim correctness are different code paths." + "reason": "Different code paths: mllama mask handling versus MPS attention correctness." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "Different tokenizer regressions: missing BOS/EOS insertion vs LlamaTokenizer overriding byte-level pre-tokenization." - }, - { - "left": "issue:41720", - "right": "issue:43540", - "accept": false, - "reason": "Qwen3 auto device mapping CUDA assert and Qwen3OmniMoe video ValueError are distinct failures." + "reason": "Separate tokenizer bugs affecting different models and different tokenizer layers." }, { "left": "issue:43540", - "right": "issue:44479", + "right": "issue:44560", "accept": false, - "reason": "Both are video-related but not the same concrete bug; one is Qwen3OmniMoe input handling, the other is a broader v5.3.0 regression across several Qwen VL models." + "reason": "Both are video-related, but they hit different Qwen model families and error types; not clearly the same bug." }, { "left": "issue:42915", - "right": "issue:43749", + "right": "issue:47332", "accept": false, - "reason": "Different loading mechanisms: FineGrainedFP8Config failure vs FSDP CPU RAM-efficient loading breakage." + "reason": "Unrelated offloading/save issue versus a FineGrainedFP8Config failure." }, { - "left": "issue:43540", - "right": "issue:44560", + "left": "issue:45042", + "right": "issue:45081", "accept": false, - "reason": "Different model families and failure modes: Qwen3OmniMoe video processing vs Qwen3-VL-embedding StopIteration." + "reason": "Completely different subsystems: image processor dependency handling versus Mistral tokenizer regex patching." }, { "left": "issue:42915", - "right": "issue:47332", + "right": "issue:43749", "accept": false, - "reason": "FineGrainedFP8Config failure and disk-offloaded save failure are unrelated offloading/quantization bugs." + "reason": "Different problems in Qwen/FP8 versus FSDP CPU RAM efficient loading." }, { "left": "issue:27957", "right": "issue:28282", "accept": false, - "reason": "XLMRoberta Flash Attention support and missing PyTorch import are unrelated." + "reason": "Unrelated issues: Flash Attention support versus missing PyTorch import error." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "gguf_file loading breakage and Apple Silicon caching_allocator_warmup TypeError are different subsystems." - }, - { - "left": "issue:45042", - "right": "issue:45081", - "accept": false, - "reason": "PIL backend dependency regression and Mistral regex patch crash are unrelated." + "reason": "Different failures in tokenizer/model loading versus Apple Silicon allocator warmup." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "TrainerState token counting and CustomTrainer.compute_loss signature mismatch are different trainer bugs." + "reason": "Trainer token-count tracking and custom compute_loss signature are distinct trainer API issues." }, { "left": "issue:44479", "right": "issue:45381", - "accept": false, - "reason": "Related video-input area, but one is a broad regression across multiple models and the other is a specific qwen2.5-vl vision_position_ids bug; not clearly the same fix." + "accept": true, + "reason": "Both describe the same Qwen2.5-VL/Qwen video-position regression in v5.3.0." }, { "left": "issue:44062", "right": "issue:45081", "accept": false, - "reason": "Tokenizer AddedToken construction error and Mistral regex patch AttributeError are unrelated." - }, - { - "left": "issue:43577", - "right": "issue:46512", - "accept": false, - "reason": "BLIP2 dtype propagation to qformer and from_config leaving config.dtype stale are different dtype-state issues." - }, - { - "left": "issue:38175", - "right": "issue:44898", - "accept": false, - "reason": "SigLIP2 zero probabilities and Perceiver non-default resolution failure are unrelated model-specific bugs." + "reason": "Tokenizer AddedToken argument conflict is unrelated to the Mistral regex/backend_tokenizer crash." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "MobileLLM loading key error and Float8 storage lookup TypeError are different failures." + "reason": "Different loading errors with no shared concrete code path." }, { "left": "issue:42175", "right": "issue:44488", "accept": false, - "reason": "Packaging/backend extras issue is unrelated to a model-loading problem for cjvt/sleng-bert." + "reason": "Package/backend installation issue versus a model loading failure." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Different models and symptoms: zero probabilities in SigLIP2 versus Perceiver non-default resolution failure." }, { "left": "issue:43493", "right": "issue:43994", "accept": true, - "reason": "Both reports describe the same SigLIP2 base-patch16-224 correctness bug: 43493 frames it as an HF-vs-JAX discrepancy, and 43994 shows the user-facing wrong outputs in AutoModel/pipeline." + "reason": "Same SigLIP2 family and same underlying implementation discrepancy producing incorrect outputs." + }, + { + "left": "issue:43577", + "right": "issue:46512", + "accept": false, + "reason": "Both involve dtype, but one is BLIP2/qformer dtype loading and the other is stale config.dtype from from_config; not the same bug." }, { "left": "issue:28282", "right": "issue:28370", "accept": false, - "reason": "Missing PyTorch import and missing vocab_file attribute for custom SentencePiece models are unrelated." + "reason": "Missing PyTorch import error and missing vocab_file for custom SentencePiece models are unrelated." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "Janus image-generation test failure and Phi-4 multimodal load failure are different model issues." + "reason": "Different multimodal model failures with no clear shared concrete code-path bug." }, { "left": "issue:43742", "right": "issue:47333", "accept": false, - "reason": "MobileLLM loading key error and disk-offloaded Qwen3 save failure are unrelated." - }, - { - "left": "issue:44987", - "right": "issue:45907", - "accept": false, - "reason": "Model loading regression for physical-intelligence/fast and list_repo_files revision forwarding are unrelated." + "reason": "Different model loading versus disk-offloaded save failure." }, { "left": "issue:44849", "right": "issue:45701", "accept": false, - "reason": "Qwen3.5 output_hidden_states bug and generic tokenization-change report are not the same issue." + "reason": "Qwen3.5 hidden-states bug and a tokenization-version change are unrelated." }, { - "left": "issue:43756", - "right": "issue:45325", + "left": "issue:44987", + "right": "issue:45907", "accept": false, - "reason": "Smollm3 RoPE-layer dropping and qwen2.5-vl temporal position scaling are different model/path bugs." + "reason": "Different repository/package issues; only loosely both mention versioning." }, { "left": "issue:43716", "right": "issue:46701", "accept": false, - "reason": "Mistral-3 image-preprocessor dtype mismatch and DiffusionGemma pipeline failure are unrelated." + "reason": "Mistral-3 image dtype mismatch and DiffusionGemma pipeline failure are unrelated." }, { - "left": "issue:45538", - "right": "issue:46869", + "left": "issue:45823", + "right": "issue:46463", + "accept": false, + "reason": "Different subsystems: Gemma4 device mismatch versus T5 error messaging." + }, + { + "left": "issue:43756", + "right": "issue:45325", "accept": false, - "reason": "CLIPTokenizer max_length regression and YOLO auto device-map failure are unrelated." + "reason": "Different Qwen/Smollm RoPE-positioning bugs with no clear shared underlying change." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The cluster summary is grounded in the packet: most pairs are clearly cross-subsystem and the only plausible duplicate signal is the SigLIP2 pair (43493/43994). The soft-edge rejections are appropriately conservative, and the one accepted pair is supported by the shared model/symptom context without overreaching." + "accept": false, + "feedback": "The cluster summary is mostly grounded, and issue:44479 is a reasonable canonical representative. However, at least one soft-edge verdict is a bit too aggressive: issue:43540 vs issue:44479 is not clearly the same bug from the packet alone. The titles show related video-input failures in different Qwen model lines, but not enough evidence to call them the same regression family with confidence. I\u2019d keep that pair conservative/uncertain rather than accept. The rest of the verdicts look more defensible." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -7973,11 +7928,11 @@ "issue:45005", "issue:45072", "issue:45405", + "issue:45588", "issue:45600", "issue:45663", "issue:45672", "issue:45674", - "issue:45823", "issue:46121", "issue:46399", "issue:46463", @@ -7993,12 +7948,11 @@ "issue:47405" ], "soft_pairs": [ - "issue:45823|issue:46463", "issue:45672|issue:45674", "issue:43756|issue:46399", "issue:46463|issue:47276", - "issue:43257|issue:43756", "issue:44737|issue:46463", + "issue:43257|issue:43756", "issue:45405|issue:45600", "issue:44625|issue:44877", "issue:43257|issue:47405", @@ -8008,222 +7962,223 @@ "issue:43906|issue:46693", "issue:44466|issue:45005", "issue:43452|issue:44488", - "issue:44568|issue:46491", "issue:24643|issue:29870", + "issue:44568|issue:46491", "issue:46463|issue:46899", - "issue:41762|issue:44936", - "issue:42796|issue:43906", "issue:44821|issue:46821", + "issue:42796|issue:43906", "issue:43452|issue:44843", + "issue:41762|issue:44936", "issue:43720|issue:44898", "issue:45663|issue:46846", "issue:43653|issue:44361", "issue:43697|issue:46829", "issue:44913|issue:46121", + "issue:44737|issue:45588", "issue:44756|issue:47404" ], - "prepared_review_unit_hash": "e7e967e76f629e11410b0acd402fb47631018e1435c043f4c08278ce1c016be2", + "prepared_review_unit_hash": "443cf992291255607b58f0457d7c1d66b31360dde4c1df90f7653674b8a9109a", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36674, - "estimated_input_tokens": 9169, - "estimated_eval_tokens": 18594 + "serialized_chars": 36702, + "estimated_input_tokens": 9176, + "estimated_eval_tokens": 18608 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of false-positive similarity matches across unrelated Transformers bugs. The only plausible duplicate pair is the v5 tied-weights/serialization regression: 44466 appears to be a specific manifestation of 45005.", - "confidence": 0.71, - "canonical_issue_reason": "issue:45005 is the broadest and most representative report of the tied-weights regression in v5; 44466 looks like a concrete device-dependent symptom of the same underlying bug.", + "summary": "The cluster is mostly heterogeneous: nearly all pairs are different bugs in unrelated models/code paths. The only strong duplicate-looking pair is the v5 tied-weights regression in translation models. No PRs are present.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 45005 is the broadest and most representative report for the only clear overlap in this set: tied-weight handling/serialization regressions in v5.", "canonical_pr_reason": null, - "best_issue_reason": "Same as canonical: 45005 is the best single issue to keep because it captures the broader tied-weights problem, while 44466 is narrower.", + "best_issue_reason": "Issue 45005 is the best global anchor because it captures the underlying tied-weights regression at a broader level than the narrower device-dependent serialization report.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:45823", - "right": "issue:46463", - "accept": false, - "reason": "Different bugs: a Gemma4 device_map forward mismatch vs. a T5 decoder-input validation message." - }, { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Both involve 4-bit inference, but they are different failure modes with different triggers and fixes." + "reason": "Both involve 4-bit/quantization, but one is a torch.finfo dtype cast bug and the other is a BitsAndBytes skip-module exclusion bug; different failure modes and code paths." }, { "left": "issue:43756", "right": "issue:46399", "accept": false, - "reason": "Both touch RoPE-related code, but one is an intended-layer-count bug and the other is a kernel registration crash." + "reason": "Different model families and defects: Smollm3 RoPE-layer dropping vs Qwen3.5 kernel registration failure." }, { "left": "issue:46463", "right": "issue:47276", "accept": false, - "reason": "Both are message-quality issues, but the affected code paths and symptoms are unrelated." + "reason": "One is T5 decoder-input validation messaging; the other is a synced_gpus warning in continuous batching. Not the same bug." }, { - "left": "issue:43257", - "right": "issue:43756", + "left": "issue:44737", + "right": "issue:46463", "accept": false, - "reason": "Different model-loading/weight-conversion bug versus a RoPE-layer-count issue." + "reason": "XLNet CPU device placement bug and T5 decoder validation error are unrelated." }, { - "left": "issue:44737", - "right": "issue:46463", + "left": "issue:43257", + "right": "issue:43756", "accept": false, - "reason": "CPU placement in XLNet positional encoding is unrelated to T5 error-message validation." + "reason": "Different problems: Qwen3 MoE weight conversion/loading with accelerate+deepspeed vs Smollm3 RoPE layer selection." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "A dependency-version bump problem and a stale auto-mapping reference are unrelated CI issues." + "reason": "PEFT version bump is a dependency/versioning issue; auto_mappings referencing removed configs is a code-maintenance breakage." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Both are config-loading problems, but they concern different models and different strictness failures." + "reason": "Qwen3.5 num_labels propagation and strict granite_speech config loading are different config-handling bugs." }, { "left": "issue:43257", "right": "issue:47405", "accept": false, - "reason": "Both involve loading/conversion, but one is missing MOE weight conversion and the other is a silent wrong-auto-class conversion skip." + "reason": "Both touch loading/conversion, but one is a Qwen3 MoE weight conversion issue and the other is a generic silent fallback when using the wrong Auto class." }, { "left": "issue:43742", "right": "issue:47332", "accept": false, - "reason": "Different loading/offloading problems with different models and failure mechanisms." + "reason": "MobileLLM loading KeyError vs disk-offloaded save failure are different offloading/loading paths." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both are dtype mismatches, but they affect different model families and code paths." + "reason": "Both are dtype-related, but the affected models and symptoms differ: BLIP2 stays float32 vs dtype mismatches during bfloat16 inference in other wrappers." }, { "left": "issue:43540", "right": "issue:46693", "accept": false, - "reason": "A multimodal input validation error is not the same as a long-context FA2 performance regression." + "reason": "Video input validation in Qwen3OmniMoe is unrelated to the FA2 long-context performance regression." }, { "left": "issue:43906", "right": "issue:46693", "accept": false, - "reason": "An isolated reproduction of another issue is not enough to match a long-context generation regression." + "reason": "An isolated reproduction of another issue is not the same as the FA2 performance regression described here." }, { "left": "issue:44466", "right": "issue:45005", "accept": true, - "reason": "Both describe the v5 tied-weights regression; 44466 is a specific device-dependent serialization symptom of the broader issue in 45005." + "reason": "Both describe the same v5 tied-weights regression: inconsistent lm_head/tied-weight serialization depending on model/device, especially for translation models." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "Both mention loading failures, but one is a gguf_file regression and the other is a specific model load problem." + "reason": "gguf_file breaks in tokenizer/model loading generally, while cjvt/sleng-bert appears to be a separate model-loading failure with no clear shared root cause." }, { - "left": "issue:44568", - "right": "issue:46491", + "left": "issue:24643", + "right": "issue:29870", "accept": false, - "reason": "Different tokenizer regressions affecting different models and round-trip behaviors." + "reason": "DeepSpeed training 'weight must be 2-D' and Bart generate decoder_attention_mask unbound local error are different failures." }, { - "left": "issue:24643", - "right": "issue:29870", + "left": "issue:44568", + "right": "issue:46491", "accept": false, - "reason": "DeepSpeed training weight-shape error and BART generate() UnboundLocalError are unrelated bugs." + "reason": "Different tokenizer regressions: missing BOS/EOS addition vs losing one leading space on round-trip." }, { "left": "issue:46463", "right": "issue:46899", "accept": false, - "reason": "An error-message bug in T5 is unrelated to Gemma4 4-bit multimodal dtype-cast failures." + "reason": "T5 decoder validation messaging and Gemma4 multimodal 4-bit dtype casts are unrelated." }, { - "left": "issue:41762", - "right": "issue:44936", + "left": "issue:44821", + "right": "issue:46821", "accept": false, - "reason": "Both are runtime/training failures, but they occur in different stages and code paths." + "reason": "AutoImageProcessor URL loading and TimesFM window_size AttributeError are different APIs and bugs." }, { "left": "issue:42796", "right": "issue:43906", "accept": false, - "reason": "The second is only an isolated reproduction reference; the titles do not establish the same tokenizer bug." + "reason": "Ministral tokenizer BPE-marker handling does not match an isolated reproduction of another issue." }, { - "left": "issue:44821", - "right": "issue:46821", + "left": "issue:43452", + "right": "issue:44843", "accept": false, - "reason": "ImageProcessor URL loading and TimesFM window_size errors are unrelated." + "reason": "gguf_file loading breakage and unconditional model_info() in mistral regex patch are different code paths." }, { - "left": "issue:43452", - "right": "issue:44843", + "left": "issue:41762", + "right": "issue:44936", "accept": false, - "reason": "One is a gguf_file loading regression; the other is offline-mode breakage from an unconditional model_info() call." + "reason": "Gemma3 ZeRO-3 loading index error and trainer.evaluate() failing after train are unrelated trainer/loading issues." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "Different model-specific inference failures with no shared concrete code-path bug." + "reason": "BitNet packed-weight unpacking during accelerate load is unrelated to Perceiver resolution/interpolation failure." }, { "left": "issue:45663", "right": "issue:46846", "accept": false, - "reason": "A Gemma4 FSDP2 shared-state bug and a Qwen3.5 tensor-parallel plan bug are distinct issues." + "reason": "Gemma4 FSDP2 shared_kv_states bug and Qwen3.5 TP plan omissions are different parallelism issues." }, { "left": "issue:43653", "right": "issue:44361", "accept": false, - "reason": "Different tokenizers and different failure modes; not the same underlying bug." + "reason": "BigBirdTokenizer special-token registration and MLukeTokenizer AttributeError are separate tokenizer defects." }, { "left": "issue:43697", "right": "issue:46829", "accept": false, - "reason": "Different models and outputs; one is an RTDetr regression, the other a VideoPrism output-field bug." + "reason": "RTDetrV2 output drift in v5 and VideoPrism hidden_states wiring are different model-output bugs." }, { "left": "issue:44913", "right": "issue:46121", "accept": false, - "reason": "Config reload defaulting and rope-parameter JSON handling are separate bugs." + "reason": "GPTNeoXConfig rotary_pct reload regression and JSON-list TypeError in rope validation are separate config/validation issues." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet device placement bug and flash_attention crash on s_aux=None do not share the same underlying problem." }, { "left": "issue:44756", "right": "issue:47404", "accept": false, - "reason": "Both involve OOMs during loading, but the underlying mechanisms are different." + "reason": "Both mention OOM during loading, but the causes differ: mmap on Strix Halo vs unbudgeted contiguous() transients in conversion-op loading." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: the only accepted soft pair is 44466/45005, and the justification stays within what the titles and pair score support. The rest of the verdicts are conservative false positives with reasons that match the issue titles. I don\u2019t see an overstatement that would warrant rejection." + "feedback": "Grounded in the packet and appropriately conservative overall. The only clearly duplicate-looking pair is issue:44466 \u2194 issue:45005, and the broader anchor choice for 45005 is well-supported by the titles. No PRs are present." }, "evaluator_used": true, "retried": false, @@ -8251,6 +8206,7 @@ "issue:43295", "issue:43525", "issue:43550", + "issue:43572", "issue:43643", "issue:43646", "issue:43653", @@ -8265,7 +8221,6 @@ "issue:44479", "issue:44493", "issue:44625", - "issue:44737", "issue:44743", "issue:44756", "issue:44811", @@ -8276,9 +8231,7 @@ "issue:45356", "issue:45357", "issue:45372", - "issue:45375", "issue:45542", - "issue:45588", "issue:45701", "issue:45704", "issue:45920", @@ -8286,224 +8239,226 @@ "issue:46170", "issue:46178", "issue:46701", + "issue:46832", "issue:46846", + "issue:46932", "issue:47000", "issue:47013", "issue:47332", "issue:47404" ], "soft_pairs": [ - "issue:44737|issue:45588", - "issue:44361|issue:44442", "issue:44493|issue:46050", + "issue:44361|issue:44442", "issue:38175|issue:43295", "issue:46846|issue:47404", "issue:45357|issue:45920", "issue:44756|issue:47332", - "issue:43688|issue:43749", "issue:43697|issue:45216", "issue:40444|issue:44811", + "issue:43688|issue:43749", "issue:44843|issue:44991", - "issue:43643|issue:45375", "issue:44857|issue:47000", "issue:45372|issue:46170", + "issue:43572|issue:43643", "issue:43761|issue:47013", "issue:43688|issue:45701", - "issue:43653|issue:43927", - "issue:43688|issue:45704", - "issue:43646|issue:43749", "issue:45542|issue:46178", + "issue:43646|issue:43749", + "issue:43653|issue:43927", "issue:43525|issue:46701", "issue:33666|issue:36010", + "issue:43688|issue:45704", + "issue:44464|issue:44743", + "issue:44479|issue:44625", "issue:43550|issue:44361", "issue:43295|issue:45356", - "issue:44464|issue:44743", - "issue:44479|issue:44625" + "issue:46832|issue:46932" ], - "prepared_review_unit_hash": "b1b401c9a20f656b16964f71d4bb6c775081ac6127f4d9325082923723eee4bc", + "prepared_review_unit_hash": "2f0061c3aae1f0abcd3f74f0f72957acca8e1843ef6052f3848c83129010a3a7", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 26, - "serialized_chars": 35223, - "estimated_input_tokens": 8806, - "estimated_eval_tokens": 17868 + "serialized_chars": 35146, + "estimated_input_tokens": 8787, + "estimated_eval_tokens": 17830 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: the soft pairs are mostly superficially similar (often same subsystem or regression language) but describe different models, code paths, and failure modes. I do not see any true duplicate PR/issue pair here.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "This cluster is a loose collection of unrelated Transformers issues, mostly centered on regressions after v5.x across tokenization, config loading, multimodal processors, and model-specific runtime bugs. None of the soft pairs looks like the same underlying bug/change, so all should stay separate.", + "confidence": 0.93, + "canonical_issue_reason": "Issue #45920 is the best representative anchor: it\u2019s a broad, concrete tokenizer regression affecting multiple models, matching the cluster\u2019s recurring theme of v5-era loading/tokenization breakage.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "#45920 is the strongest standalone issue in the set because it is clearly described, widely scoped, and central to the shared regression theme, making it the best canonical issue if one must be chosen.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44737", - "right": "issue:45588", + "left": "issue:44493", + "right": "issue:46050", "accept": false, - "reason": "Both are runtime/perf bugs, but one is XLNet CPU device placement and the other is a flash-attention crash on sink-less models; different code paths and fixes." + "reason": "Different failures and code paths: a position-id key warning across many models vs a quantized model save/load corruption bug." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "Both are tokenizer-related, but MLukeTokenizer task failures and AutoTokenizer loading FastSpeech2ConformerTokenizer are distinct loader/behavior bugs." - }, - { - "left": "issue:44493", - "right": "issue:46050", - "accept": false, - "reason": "One is a position-id key regression across models; the other is save/load breakage for sinq-quantized models. Different functionality and root cause." + "reason": "Both are tokenizer-related, but they involve different tokenizer classes and distinct loading/attribute errors, not the same bug." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "Siglip2 zero probabilities and processor.tokenizer/image-passing regression are unrelated model behavior vs processor API breakage." + "reason": "Unrelated symptoms: zero probabilities in SigLIP2 inference vs a regression in processor.tokenizer / image-to-tokenizer handling." }, { "left": "issue:46846", "right": "issue:47404", "accept": false, - "reason": "Qwen3.5 tensor-parallel planning and device_map loading OOM from conversion transients are different memory-planning issues in different code paths." + "reason": "Both mention OOM, but one is tensor-parallel plan sizing and the other is device_map loading transients; different mechanisms and fixes." }, { "left": "issue:45357", "right": "issue:45920", "accept": false, - "reason": "Both involve Qwen/Tokenizer regressions, but one is Qwen3.5 save_pretrained visual encoder keys and the other is broad token ID errors across many models; not the same bug." + "reason": "Different layers of the stack: Qwen3.5 save_pretrained visual-key serialization vs a tokenizer ID regression across multiple models." }, { "left": "issue:44756", "right": "issue:47332", "accept": false, - "reason": "Both mention offloading/memory, but Strix Halo mmap OOM and inability to save disk-offloaded Qwen3-VL-32B are different storage/offload problems." - }, - { - "left": "issue:43688", - "right": "issue:43749", - "accept": false, - "reason": "Auxiliary loss normalization in OLMoE/GPT OSS is unrelated to FSDP_CPU_RAM_EFFICIENT_LOADING breakage." + "reason": "Different offloading problems: mmap avoidance on Strix Halo to prevent OOM vs inability to save a disk-offloaded Qwen3-VL checkpoint." }, { "left": "issue:43697", "right": "issue:45216", "accept": false, - "reason": "RTDetrV2 output drift in v5 and Qwen3.5 save_pretrained regression affect different models and different paths (inference vs checkpoint saving)." + "reason": "Separate model-specific regressions: RT-DETR output differences vs Qwen3.5 checkpoint save_pretrained corruption." }, { "left": "issue:40444", "right": "issue:44811", "accept": false, - "reason": "IterableDataset multi-image finetuning failure is unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "One is finetuning with multiple images per prompt; the other is Whisper batch_decode skip_special_tokens handling. Not the same bug." }, { - "left": "issue:44843", - "right": "issue:44991", + "left": "issue:43688", + "right": "issue:43749", "accept": false, - "reason": "Both touch tokenizer loading, but offline hub model_info() calls and EMBEDDIA/est-roberta tokenizer loading failure are different bugs with different triggers." + "reason": "Auxiliary-loss normalization in MoE models is unrelated to FSDP CPU RAM efficient loading." }, { - "left": "issue:43643", - "right": "issue:45375", + "left": "issue:44843", + "right": "issue:44991", "accept": false, - "reason": "AutoConfig trust_remote_code missing fields is a generic config-loading issue; Qwen3_5MoeVisionConfig missing deepstack_visual_indexes is a specific schema field omission." + "reason": "Both are tokenizer load regressions, but one is an offline hub call in _patch_mistral_regex and the other is a specific model tokenizer load failure; no clear same code path." }, { "left": "issue:44857", "right": "issue:47000", "accept": false, - "reason": "LwDetrImageLoss AMP crash and RTDetrLoss infeasible cost matrix are different detection-loss failures with different failure points." + "reason": "Both concern detection losses, but one crashes under float16 AMP/CUDA and the other has an infeasible cost matrix in RTDetrLoss; different bugs." }, { "left": "issue:45372", "right": "issue:46170", "accept": false, - "reason": "Gemma 4 processor ImportError from mistral_common and Windows ROCm torch.distributed import/type error are unrelated dependency issues." + "reason": "Completely different import issues: mistral_common ReasoningEffort for Gemma 4 vs Windows/ROCm torch.distributed dependency." + }, + { + "left": "issue:43572", + "right": "issue:43643", + "accept": false, + "reason": "Both are about missing config fields, but one is a specific StableLm pad token field after v5 update and the other is a trust_remote_code AutoConfig field-loss issue." }, { "left": "issue:43761", "right": "issue:47013", "accept": false, - "reason": "CLIPVisionModel hidden_states regression and xLSTM train-mode unpacking crash are separate model implementations with different symptoms." + "reason": "Different model families and mechanics: CLIPVision hidden_states regression vs xLSTM tuple-unpacking crash with return_last_states=False." }, { "left": "issue:43688", "right": "issue:45701", "accept": false, - "reason": "Auxiliary loss normalization bug is unrelated to version-dependent tokenization changes." - }, - { - "left": "issue:43653", - "right": "issue:43927", - "accept": false, - "reason": "BigBirdTokenizer special-token decode bug and DiaConfig losing custom token IDs are both token/config issues, but not the same defect or fix." + "reason": "Auxiliary-loss normalization is not the same as a broad tokenization behavior change report." }, { - "left": "issue:43688", - "right": "issue:45704", + "left": "issue:45542", + "right": "issue:46178", "accept": false, - "reason": "Auxiliary loss normalization in MoE models is unrelated to T5 using apex.FusedRMSNorm memory leak." + "reason": "Both are backend-related, but tensorboard/TensorFlow backend detection is separate from the tensorflow_text backend mapping error." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Custom model initialization breakage in Transformers 5.0 and FSDP CPU RAM efficient loading failure are different regressions." + "reason": "General custom-model initialization breakage is unrelated to FSDP CPU RAM efficient loading." }, { - "left": "issue:45542", - "right": "issue:46178", + "left": "issue:43653", + "right": "issue:43927", "accept": false, - "reason": "TensorBoard-only backend error without TensorFlow and tensorflow_text backend mapping error are different backend-registration issues." + "reason": "BigBirdTokenizer special-token registration and DiaConfig custom token ID persistence are different classes of bugs." }, { "left": "issue:43525", "right": "issue:46701", "accept": false, - "reason": "Llama4Config missing pad_token_id and DiffusionGemma image-text-to-text pipeline failure are unrelated model/config problems." + "reason": "A missing pad_token_id on Llama4Config is unrelated to the DiffusionGemma image-text-to-text pipeline failure." }, { "left": "issue:33666", "right": "issue:36010", "accept": false, - "reason": "Qwen2-VL multi-GPU training and GenerationMixin ImportError are completely different issues." + "reason": "Multi-GPU Qwen2-VL training and a GenerationMixin import error are unrelated issues." }, { - "left": "issue:43550", - "right": "issue:44361", - "accept": false, - "reason": "Bamba-9B-v2 torch.compile SDPA failure and MLukeTokenizer AttributeError target different models and code paths." - }, - { - "left": "issue:43295", - "right": "issue:45356", + "left": "issue:43688", + "right": "issue:45704", "accept": false, - "reason": "Processor.tokenizer regression and Kimi-K2.5 tokenizer codec/regex regression are both tokenizer-related but distinct breakages." + "reason": "MoE auxiliary-loss normalization is unrelated to T5 silently using apex.FusedRMSNorm and leaking memory." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward and Qwen3_5 recurrent-state reset in cache are separate generation-state bugs." + "reason": "Both involve generation/state, but one is compiled-forward chunked generation inconsistency and the other is xLSTM recurrent-state reset logic; not the same fix." }, { "left": "issue:44479", "right": "issue:44625", "accept": false, - "reason": "Video input regression for Qwen VL models and Qwen3.5 num_labels propagation are different Qwen family issues with different affected features." + "reason": "A multimodal video-input regression in several Qwen models is distinct from Qwen3.5 num_labels propagation." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "torch.compile/SDPA on Bamba-9B-v2 is unrelated to MLukeTokenizer task-time AttributeError." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Processor/tokenizer API regression with images is unrelated to Kimi-K2.5 codec handling and mistral regex warning behavior." + }, + { + "left": "issue:46832", + "right": "issue:46932", + "accept": false, + "reason": "RT-DETR forward shape/type error is unrelated to Gemma4 quantized vision/audio embedder dtype casting." } ] }, @@ -8542,6 +8497,7 @@ "issue:43572", "issue:43577", "issue:43606", + "issue:43611", "issue:43643", "issue:43653", "issue:43688", @@ -8553,269 +8509,267 @@ "issue:44361", "issue:44464", "issue:44466", - "issue:44484", "issue:44554", "issue:44561", "issue:44568", "issue:44610", "issue:44625", - "issue:44704", "issue:44734", "issue:44964", "issue:44977", "issue:45405", "issue:45440", + "issue:45823", "issue:46050", "issue:46402", "issue:46491", "issue:46612", "issue:46693", "issue:46829", - "issue:46832", - "issue:46932", + "issue:46846", "issue:47254", "issue:47276" ], "soft_pairs": [ "issue:43756|issue:46402", - "issue:46832|issue:46932", - "issue:43653|issue:44568", "issue:43761|issue:46829", + "issue:43653|issue:44568", "issue:41720|issue:44464", - "issue:43452|issue:43577", "issue:33666|issue:36331", + "issue:43452|issue:43577", "issue:28370|issue:34567", - "issue:43504|issue:43550", "issue:44466|issue:46050", - "issue:44464|issue:44977", "issue:44977|issue:46693", + "issue:43504|issue:43550", + "issue:44464|issue:44977", "issue:32090|issue:33666", "issue:42898|issue:46491", + "issue:45405|issue:47254", "issue:43606|issue:43927", "issue:43720|issue:43927", - "issue:43504|issue:43720", - "issue:45405|issue:47254", "issue:43782|issue:46693", + "issue:43504|issue:43720", "issue:43278|issue:43688", "issue:43643|issue:44561", + "issue:44625|issue:47276", "issue:43720|issue:44361", - "issue:43688|issue:45440", "issue:43504|issue:44610", - "issue:44625|issue:47276", + "issue:43688|issue:45440", "issue:44734|issue:46612", - "issue:43577|issue:44554", - "issue:44484|issue:44704", "issue:43525|issue:43572", - "issue:44964|issue:45405" + "issue:43577|issue:44554", + "issue:44964|issue:45405", + "issue:45823|issue:46050", + "issue:43611|issue:46846" ], - "prepared_review_unit_hash": "b9523a79e289de9d54e3063aa83c9f703bdc8f935a92c38e9bc9735bc39fd7ae", + "prepared_review_unit_hash": "401ab4d89f30dc340aee34b7e3578c004c429ad5851ab7fd947c15bd5d82c3f7", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36850, - "estimated_input_tokens": 9213, - "estimated_eval_tokens": 18682 + "serialized_chars": 36299, + "estimated_input_tokens": 9075, + "estimated_eval_tokens": 18406 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No soft-edge pair looks like the same underlying bug/change; the set is a grab-bag of unrelated model-, tokenizer-, training-, and generation-regression reports. I would not merge any of the proposed pairs.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue stands out because the candidates are not duplicates of one another; they cover distinct subsystems and model-specific failures.", + "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer, config, loading, training, generation, and serving bugs. None of the soft pairs looks like the same underlying fix, so no duplicate merge is justified.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits the cluster because the items span unrelated regressions and model families rather than one shared bug.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46693 is the strongest representative issue here: it is an open, well-scoped regression report with broad impact, clear reproduction, and high discussion activity.", + "best_issue_reason": "No issue is a good global representative; the set is too broad and the reported failures are unrelated in codepath and fix.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43756", "right": "issue:46402", "accept": false, - "reason": "Different models and failure modes: RoPE-layer loss vs. text-encoder weights not loading." + "reason": "Different models and different failures: RoPE layer dropping in Smollm3 versus text encoder weights not loading in Sam3Model." }, { - "left": "issue:46832", - "right": "issue:46932", + "left": "issue:43761", + "right": "issue:46829", "accept": false, - "reason": "Unrelated bugs: RT-DETR feature-level mismatch vs. Gemma4 quantized embedder dtype casting." + "reason": "Both mention hidden states, but they are separate model-specific output bugs with different APIs and symptoms." }, { "left": "issue:43653", "right": "issue:44568", "accept": false, - "reason": "Both tokenizer/special-token related, but different tokenizers and different symptoms; not the same bug." - }, - { - "left": "issue:43761", - "right": "issue:46829", - "accept": false, - "reason": "Different output-contract issues in different vision models; hidden_states missing vs misassigned." + "reason": "Tokenizer special-token registration for BigBird is not the same bug as BOS/EOS insertion behavior for mdeberta-v3." }, { "left": "issue:41720", "right": "issue:44464", "accept": false, - "reason": "Different code paths: device-mapping CUDA assert vs. compiled-forward chunked generation inconsistency." + "reason": "CUDA assert during Qwen3 device mapping is unrelated to chunked generation inconsistencies under compiled forward." }, { - "left": "issue:43452", - "right": "issue:43577", + "left": "issue:33666", + "right": "issue:36331", "accept": false, - "reason": "Different regressions: gguf tokenizer/model loading vs. dtype mismatch in BLIP-2 loading." + "reason": "Multi-GPU training on Qwen2-VL is not the same as a Trainer API kwarg change breaking custom compute_loss." }, { - "left": "issue:33666", - "right": "issue:36331", + "left": "issue:43452", + "right": "issue:43577", "accept": false, - "reason": "Both involve training, but one is multi-GPU broadcasting and the other is a custom compute_loss signature break." + "reason": "gguf_file loading breakage and BLIP2 dtype persistence are different loading paths with unrelated root causes." }, { "left": "issue:28370", "right": "issue:34567", "accept": false, - "reason": "Unrelated: missing vocab_file on custom SentencePiece models vs. TrainerState token counter not updating." + "reason": "Missing vocab_file in custom SentencePiece models is unrelated to TrainerState.num_input_tokens_seen not updating." }, { - "left": "issue:43504", - "right": "issue:43550", + "left": "issue:44466", + "right": "issue:46050", "accept": false, - "reason": "Different subsystems and symptoms: legacy config field on BEiT vs. torch.compile/SDPA failure in Bamba." + "reason": "Serialization of tied lm_head weights is a different bug from quantized sinq save/load corruption." }, { - "left": "issue:44466", - "right": "issue:46050", + "left": "issue:44977", + "right": "issue:46693", "accept": false, - "reason": "Both touch save/load behavior, but one is tied-weight serialization and the other is quantized sinq persistence." + "reason": "Flash-attention generation correctness on Qwen3.5 is not the same underlying issue as a long-context FA2 performance regression." }, { - "left": "issue:44464", - "right": "issue:44977", + "left": "issue:43504", + "right": "issue:43550", "accept": false, - "reason": "Different generation bugs with different triggers: compiled forward inconsistency vs. Qwen3.5 flash-attention failure." + "reason": "Legacy-field model loading in Beit is unrelated to a torch.compile + SDPA failure in Bamba." }, { - "left": "issue:44977", - "right": "issue:46693", + "left": "issue:44464", + "right": "issue:44977", "accept": false, - "reason": "Both mention FA2, but one is a model-specific correctness issue and the other is a long-context performance regression." + "reason": "Chunked generation with compiled forward and flash-attention generation failures are distinct codepaths and fixes." }, { "left": "issue:32090", "right": "issue:33666", "accept": false, - "reason": "Both training-related, but the errors are different and not the same multi-GPU bug." + "reason": "A NoneType GPU broadcast error in Trainer is not the same as Qwen2-VL multi-GPU training behavior." }, { "left": "issue:42898", "right": "issue:46491", "accept": false, - "reason": "General tokenization-space behavior change vs. a CodeLlama leading-space round-trip regression." + "reason": "Global clean_up_tokenization_spaces behavior changes are unrelated to CodeLlama's leading-space round-trip regression." }, { - "left": "issue:43606", - "right": "issue:43927", + "left": "issue:45405", + "right": "issue:47254", "accept": false, - "reason": "Device-mismatch with CPU offload is unrelated to DiaConfig losing custom token IDs on save/load." + "reason": "A release/version pin issue for PEFT is not the same as a runtime CheckpointError with ZeRO-3 and gradient checkpointing." }, { - "left": "issue:43720", + "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "Different failure classes: packed-weight unloading during accelerate load vs. config field loss." + "reason": "CPU offload device mismatch in Bark is unrelated to DiaConfig losing custom token IDs on save/load." }, { - "left": "issue:43504", - "right": "issue:43720", + "left": "issue:43720", + "right": "issue:43927", "accept": false, - "reason": "Different models and root causes; BEiT legacy field handling is not the BitNet packed-weight issue." + "reason": "Packed-weight unpacking during accelerate loading is a different failure mode from config field loss after serialization." }, { - "left": "issue:45405", - "right": "issue:47254", + "left": "issue:43782", + "right": "issue:46693", "accept": false, - "reason": "One is a dependency-version bump concern; the other is a runtime checkpointing error with PEFT/ZeRO-3." + "reason": "Qwen3VL weight_only loading error is unrelated to the FA2 long-context performance regression." }, { - "left": "issue:43782", - "right": "issue:46693", + "left": "issue:43504", + "right": "issue:43720", "accept": false, - "reason": "Different symptoms and code paths: Qwen3VL weight_only load error vs. FA2 long-context performance regression." + "reason": "A legacy field blocking Beit preset loading is not the same as BitNet packed-weight handling in accelerate." }, { "left": "issue:43278", "right": "issue:43688", "accept": false, - "reason": "Different numerical issues in different models: evaluation dtype drift vs. auxiliary-loss normalization." + "reason": "Embedding dtype drift between train/eval and auxiliary loss normalization are separate numerical issues." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Distinct trust_remote_code regressions with different missing APIs/fields and different breakage mechanisms." + "reason": "Both involve trust_remote_code, but one is missing fields in AutoConfig output while the other is an API removal breaking remote models." }, { - "left": "issue:43720", - "right": "issue:44361", + "left": "issue:44625", + "right": "issue:47276", "accept": false, - "reason": "Packed-weight loading bug is unrelated to MLukeTokenizer AttributeError on tasks." + "reason": "Qwen3.5 num_labels propagation is unrelated to a misleading synced_gpus warning message." }, { - "left": "issue:43688", - "right": "issue:45440", + "left": "issue:43720", + "right": "issue:44361", "accept": false, - "reason": "Both are model correctness issues, but they affect different MoE implementations and different loss/forward logic." + "reason": "BitNet packed-weight loading and MLukeTokenizer AttributeError affect different subsystems and codepaths." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "Unrelated: BEiT legacy config field vs. OmDet-Turbo processor/image-size mismatch." + "reason": "Legacy-field loading in Beit is unrelated to OmDet-Turbo's processor/model input-size mismatch." }, { - "left": "issue:44625", - "right": "issue:47276", + "left": "issue:43688", + "right": "issue:45440", "accept": false, - "reason": "Different problems: config propagation for Qwen3.5 vs. a misleading warning in continuous batching." + "reason": "Auxiliary loss normalization is a different bug from DeepseekV3 implementation divergence." }, { "left": "issue:44734", "right": "issue:46612", "accept": false, - "reason": "Both involve generation internals, but one is KV-cache continuation indexing and the other is beam-search cache reorder omissions." + "reason": "KV-cache continuation indexing in serve is not the same as beam-search cache reorder being skipped for specific models." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config and missing pad_token_idx in StableLmConfig are separate config-field regressions." }, { "left": "issue:43577", "right": "issue:44554", "accept": false, - "reason": "Different issues in different execution environments: BLIP-2 dtype loading vs. MPS attention correctness." + "reason": "BLIP2 dtype persistence is unrelated to an MPS attention correctness issue when value head dim differs from query." }, { - "left": "issue:44484", - "right": "issue:44704", + "left": "issue:44964", + "right": "issue:45405", "accept": false, - "reason": "Save_pretrained shard-size question is unrelated to AutoProcessor kwargs propagation." + "reason": "Phi-4 multimodal loading failure and unreleased PEFT version pin are unrelated." }, { - "left": "issue:43525", - "right": "issue:43572", + "left": "issue:45823", + "right": "issue:46050", "accept": false, - "reason": "Similar class of config-field regressions, but different models and different missing fields; not the same bug." + "reason": "Gemma4 PLE device mismatch during forward is a different bug from quantized sinq save/load breakage." }, { - "left": "issue:44964", - "right": "issue:45405", + "left": "issue:43611", + "right": "issue:46846", "accept": false, - "reason": "Model-loading failure for Phi-4 multimodal is unrelated to the PEFT minimum-version bump." + "reason": "base_model_prefix load regression and missing linear_attn entries in base_model_tp_plan are separate model-loading and tensor-parallel planning issues." } ] }, @@ -8853,7 +8807,6 @@ "issue:43531", "issue:43572", "issue:43606", - "issue:43611", "issue:43646", "issue:43701", "issue:43825", @@ -8864,255 +8817,256 @@ "issue:44291", "issue:44361", "issue:44466", + "issue:44484", "issue:44488", "issue:44514", "issue:44554", "issue:44568", + "issue:44589", + "issue:44704", "issue:44843", "issue:44977", "issue:45290", "issue:45405", "issue:45491", + "issue:45520", "issue:45759", "issue:45812", - "issue:45823", "issue:46002", - "issue:46050", "issue:46192", "issue:46209", "issue:46365", "issue:46489", - "issue:46846", "issue:47332", "issue:47404", "issue:47436" ], "soft_pairs": [ - "issue:45823|issue:46050", - "issue:28218|issue:33357", - "issue:43611|issue:46846", "issue:43334|issue:45759", + "issue:28218|issue:33357", + "issue:44484|issue:44704", "issue:44514|issue:45290", "issue:39692|issue:43295", "issue:43334|issue:43572", - "issue:43994|issue:46192", - "issue:44977|issue:46209", "issue:45405|issue:47332", + "issue:43994|issue:46192", "issue:44488|issue:44843", + "issue:44977|issue:46209", "issue:43257|issue:47404", - "issue:42796|issue:44977", "issue:43493|issue:43825", - "issue:43927|issue:44361", + "issue:42796|issue:44977", "issue:43531|issue:47436", + "issue:43927|issue:44361", "issue:43646|issue:44466", "issue:43572|issue:44291", - "issue:28218|issue:29925", "issue:45491|issue:46002", - "issue:44554|issue:46365", "issue:43278|issue:43701", - "issue:44568|issue:46489", + "issue:44554|issue:46365", + "issue:28218|issue:29925", "issue:44568|issue:45812", - "issue:43504|issue:43606", + "issue:44568|issue:46489", "issue:43873|issue:47404", + "issue:43504|issue:43606", + "issue:44589|issue:45520", "issue:43278|issue:43844", "issue:43493|issue:43844" ], - "prepared_review_unit_hash": "25caa7f687318f92bc6108d4d7ee68f7844de9b0ee77c366e77aa37a8731d2bf", + "prepared_review_unit_hash": "496cd4db430b1a80ba16274a700d84b392058db5ff6b5908630976c0b09b4547", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35772, - "estimated_input_tokens": 8943, - "estimated_eval_tokens": 18142 + "serialized_chars": 35713, + "estimated_input_tokens": 8929, + "estimated_eval_tokens": 18114 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a loose similarity cluster, not a true duplicate set: the items mostly cover unrelated tokenizer regressions, model-loading/config issues, offloading/quantization bugs, and docs/examples problems.", - "confidence": 0.95, - "canonical_issue_reason": "No clear canonical issue: the set spans many distinct bug families and model-specific regressions rather than one underlying defect.", + "summary": "The soft links are mostly false positives across unrelated transformer bugs. I would anchor the cluster on the broad, high-activity offloading/quantization report, and I do not see any pair here that is a true duplicate or same-code-path change.", + "confidence": 0.72, + "canonical_issue_reason": "Issue 43873 is the most active and broadly phrased report in the set (highest discussion/inbound activity, still open), so it is the best anchor for triage.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good representative for deduping; any choice would hide major topic differences across the cluster.", + "best_issue_reason": "Issue 43873 is the strongest standalone bug report here: high signal, broad impact, and enough activity to serve as the representative issue.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45823", - "right": "issue:46050", + "left": "issue:43334", + "right": "issue:45759", "accept": false, - "reason": "Different problems: Gemma4 device mismatch during forward vs a quantized model save/load corruption issue." + "reason": "Different failures in Qwen multimodal config handling: missing pad_token_id vs unwrapping text_config in from_config." }, { "left": "issue:28218", "right": "issue:33357", "accept": false, - "reason": "Tokenizer spacing bug vs MacOS bus error on a CLIP model; unrelated failure modes." - }, - { - "left": "issue:43611", - "right": "issue:46846", - "accept": false, - "reason": "Both are loading-related, but one is about base_model_prefix handling and the other about missing TP-plan entries for Qwen3.5; different code paths." + "reason": "Tokenizer spacing after added tokens is unrelated to a Mac bus error in a CLIP model." }, { - "left": "issue:43334", - "right": "issue:45759", + "left": "issue:44484", + "right": "issue:44704", "accept": false, - "reason": "Both involve Qwen multimodal configs, but one is a missing pad_token_id attribute and the other is from_config not unwrapping text_config." + "reason": "save_pretrained shard-size question and AutoProcessor kwargs passthrough are different issues." }, { "left": "issue:44514", "right": "issue:45290", "accept": false, - "reason": "Same API area, but different crashes: batched chat template with padding=False vs assistant messages with tool calls and no content." + "reason": "Both hit apply_chat_template, but the triggers and affected inputs differ materially." }, { "left": "issue:39692", "right": "issue:43295", "accept": false, - "reason": "Docs/example mismatch and quantization failure vs processor.tokenizer regression; not the same bug." + "reason": "Docs example mistakes are not the same bug as a processor.tokenizer/images regression." }, { "left": "issue:43334", "right": "issue:43572", "accept": false, - "reason": "Different missing-config-field bugs in different model families." - }, - { - "left": "issue:43994", - "right": "issue:46192", - "accept": false, - "reason": "SigLIP2 wrong outputs vs return_tensors being ignored; distinct tokenizer/processor issues." + "reason": "Different model families and different missing config fields." }, { - "left": "issue:44977", - "right": "issue:46209", + "left": "issue:45405", + "right": "issue:47332", "accept": false, - "reason": "Qwen3.5 flash-attention generation bug vs FP8 inference failure on B200; unrelated issues." + "reason": "PEFT version bump issue and disk-offload save failure are unrelated." }, { - "left": "issue:45405", - "right": "issue:47332", + "left": "issue:43994", + "right": "issue:46192", "accept": false, - "reason": "PEFT version pinning problem vs disk-offloaded Qwen3-VL save failure; no shared underlying defect." + "reason": "SigLIP2 output quality regression is not the same as silently ignored return_tensors kwargs." }, { "left": "issue:44488", "right": "issue:44843", "accept": false, - "reason": "Model loading failure for sleng-bert vs offline hub access bug in mistral regex patch." + "reason": "A model-loading failure and an offline-mode network call in tokenizer patching are distinct code paths." }, { - "left": "issue:43257", - "right": "issue:47404", + "left": "issue:44977", + "right": "issue:46209", "accept": false, - "reason": "MoE weight conversion with accelerate+deepspeed vs OOM from conversion-op transients in device_map loading." + "reason": "Flash-attention generation problems are not the same as FP8/B200 inference failures." }, { - "left": "issue:42796", - "right": "issue:44977", + "left": "issue:43257", + "right": "issue:47404", "accept": false, - "reason": "Tokenizer BPE marker handling vs Qwen3.5 flash-attention generation; unrelated." + "reason": "Qwen3 MoE weight conversion under accelerate+deepspeed is different from device_map transient OOMs." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 implementation discrepancy vs pipeline error-message wording about translation tasks." + "reason": "Model implementation discrepancy and pipeline error-message wording are unrelated." }, { - "left": "issue:43927", - "right": "issue:44361", + "left": "issue:42796", + "right": "issue:44977", "accept": false, - "reason": "Save/load custom token IDs loss vs MLukeTokenizer AttributeError; different tokenizer defects." + "reason": "Tokenizer BPE-marker handling does not match a Qwen3.5 flash-attention generation bug." }, { "left": "issue:43531", "right": "issue:47436", "accept": false, - "reason": "Qwen3-MoE sliding_window issue vs Qwen2_5OmniConfig missing max_position_embeddings; unrelated config/model bugs." + "reason": "Sliding-window behavior in Qwen3-MoE is not the same as a missing max_position_embeddings attribute." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "save/load token-id loss and tokenizer AttributeError on tasks are different defects." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model initialization regression vs lm_head.weight serialization inconsistency; different code paths." + "reason": "Custom model initialization breakage is distinct from tied-weight serialization varying by device." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig vs init_empty_weights unexpected argument; not the same issue." + "reason": "A missing pad_token_idx field and an init_empty_weights argument mismatch are unrelated." }, { - "left": "issue:28218", - "right": "issue:29925", + "left": "issue:45491", + "right": "issue:46002", "accept": false, - "reason": "Both tokenizer-related, but one is an added-token spacing regression and the other is a use_fast enforcement problem." + "reason": "Gemma3 sliding-window NaNs and classification-head NaNs after checkout/init are different symptoms with different causes." }, { - "left": "issue:45491", - "right": "issue:46002", + "left": "issue:43278", + "right": "issue:43701", "accept": false, - "reason": "Gemma3 mixed-length NaN embeddings vs NaNs in classification heads after checkout/init; different symptoms and likely causes." + "reason": "Embedding dtype drift is not the same as resume_from_checkpoint key mismatch." }, { "left": "issue:44554", "right": "issue:46365", "accept": false, - "reason": "Attention correctness on MPS vs importing Trainer being blocked by datasets; unrelated." + "reason": "MPS attention correctness and Trainer import conflict with datasets are unrelated." }, { - "left": "issue:43278", - "right": "issue:43701", + "left": "issue:28218", + "right": "issue:29925", "accept": false, - "reason": "Embedding dtype regression in evaluate vs resume_from_checkpoint key mismatch." + "reason": "Extra space after added token and SigLIP fast-tokenizer enforcement do not indicate the same bug." }, { "left": "issue:44568", - "right": "issue:46489", + "right": "issue:45812", "accept": false, - "reason": "Both tokenizer regressions in v5, but different models and failures: missing BOS/EOS addition vs incorrect DeepSeek-Coder output." + "reason": "mDeBERTa special-token regression and Granite token-id regression are separate tokenizer bugs." }, { "left": "issue:44568", - "right": "issue:45812", + "right": "issue:46489", + "accept": false, + "reason": "Different models, different symptoms, and no concrete shared code path are shown." + }, + { + "left": "issue:43873", + "right": "issue:47404", "accept": false, - "reason": "Different tokenizer regressions affecting different model families and different tokenization behavior." + "reason": "Quantization/offloading not working as expected is not the same as loader transient OOM from conversion ops." }, { "left": "issue:43504", "right": "issue:43606", "accept": false, - "reason": "Legacy-field model loading bug vs CPU-offload device mismatch; unrelated." + "reason": "Legacy-field model loading failure and CPU offload device mismatch are different issues." }, { - "left": "issue:43873", - "right": "issue:47404", + "left": "issue:44589", + "right": "issue:45520", "accept": false, - "reason": "Both relate to offloading/quantization, but one is quantization offload behavior and the other is transient OOM during device_map loading." + "reason": "Float8 storage lookup failure and flash_attn import KeyError are unrelated." }, { "left": "issue:43278", "right": "issue:43844", "accept": false, - "reason": "Training/eval dtype drift vs ZeRO-3 gradient growth; not the same defect." + "reason": "Embedding dtype drift does not match a ZeRO-3 gradient growth bug." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "SigLIP2 implementation mismatch vs DeepSpeed gradient blow-up; unrelated." + "reason": "SigLIP2 HF-vs-JAX discrepancy is unrelated to gradient instability under HfDeepSpeedConfig." } ] }, @@ -9186,238 +9140,231 @@ "issue:47436" ], "soft_pairs": [ - "issue:44589|issue:45520", "issue:46821|issue:46832", "issue:43824|issue:44589", "issue:46209|issue:46710", - "issue:43504|issue:43653", "issue:42490|issue:43696", + "issue:43504|issue:43653", "issue:43606|issue:44112", "issue:43994|issue:44493", "issue:44493|issue:46285", - "issue:45072|issue:46832", - "issue:44464|issue:45584", "issue:43572|issue:47436", + "issue:45072|issue:46832", "issue:43873|issue:46050", + "issue:44464|issue:45584", "issue:15354|issue:38175", "issue:45245|issue:46710", "issue:47013|issue:47254", "issue:44849|issue:45357", "issue:45084|issue:45520", - "issue:45229|issue:45823", "issue:42947|issue:43381", - "issue:28218|issue:34567", + "issue:45229|issue:45823", "issue:44423|issue:44734", - "issue:43844|issue:44964", "issue:44849|issue:46209", - "issue:45200|issue:45229", + "issue:43844|issue:44964", + "issue:28218|issue:34567", "issue:45588|issue:46050", + "issue:45200|issue:45229", "issue:37428|issue:43957", "issue:43493|issue:43701", "issue:41720|issue:46710" ], - "prepared_review_unit_hash": "24d6e29a66e93ef8d6ce03a4b2d857b5f8bbaeb54f044f6dc26bccd4df2a396b", + "prepared_review_unit_hash": "7a02a1f3c927689686e84c948dc0131a687e9049e9bf0cebfa740fa6941a31b0", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36505, - "estimated_input_tokens": 9127, - "estimated_eval_tokens": 18510 + "soft_pair_count": 28, + "serialized_chars": 36209, + "estimated_input_tokens": 9053, + "estimated_eval_tokens": 18362 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer, config, loading, checkpointing, serving, quantization, and model-specific bugs. None of the soft-similarity pairs look like the same underlying issue, so no canonical duplicate should be selected.", - "confidence": 0.98, - "canonical_issue_reason": "No single canonical issue fits; the items are about different code paths and failure modes rather than one duplicated bug.", + "summary": "The items are a heterogeneous set of unrelated issues, not a duplicate cluster. None of the soft pairs describe the same underlying bug or change closely enough to merge/close as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": "If a representative issue must be chosen, issue #43873 is the strongest single candidate because it has the richest discussion and references, but the overall set is too diverse to justify a true canonical duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong global representative because the cluster is not a duplicate set and spans many unrelated topics.", + "best_issue_reason": "Issue #43873 is the best representative of the set only in the sense of being the most substantial/actively discussed issue; it is not a duplicate anchor for the other items.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44589", - "right": "issue:45520", - "accept": false, - "reason": "Different failures: FP8 storage lookup error vs flash_attn import key error on Python 3.13." - }, { "left": "issue:46821", "right": "issue:46832", "accept": false, - "reason": "Unrelated model-specific problems: TimesFM window_size attribute error vs RT-DETR feature-level forward mismatch." + "reason": "Both are runtime errors in model code, but they affect different models, different inputs, and different failure paths." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "Different areas and symptoms: Qwen2.5-VL import error vs missing Float8 storage backend." + "reason": "An import error for a missing Qwen class and a storage-object CUDA type error are unrelated bugs." }, { "left": "issue:46209", "right": "issue:46710", "accept": false, - "reason": "FP8 inference on B200 is unrelated to an accuracy regression after a transformers upgrade." + "reason": "FP8 inference on B200 and an accuracy regression after upgrade are different problems with different symptoms and likely causes." }, { - "left": "issue:43504", - "right": "issue:43653", + "left": "issue:42490", + "right": "issue:43696", "accept": false, - "reason": "Different subsystems: BEiT pretrained loading legacy field vs BigBirdTokenizer special-token registration." + "reason": "A local-path import warning and a GPU out-of-memory error are not the same underlying issue." }, { - "left": "issue:42490", - "right": "issue:43696", + "left": "issue:43504", + "right": "issue:43653", "accept": false, - "reason": "Local-path import warning is unrelated to GPT-oss CUDA out-of-memory." + "reason": "Loading a BEiT preset with a legacy field and BigBird tokenizer special-token registration are unrelated model/tokenizer bugs." }, { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "One is a real CPU offload device mismatch; the other is a stale CI test for GraniteSpeech." + "reason": "A real CPU-offload device mismatch in Bark is not the same as a stale CI test failure in GraniteSpeech." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "Both are model output issues, but the concrete bugs differ: SigLIP2 bad outputs vs unexpected position-id key handling." + "reason": "SigLIP2 producing nonsense outputs and a widespread position-id key warning point to different code paths and failure modes." }, { "left": "issue:44493", "right": "issue:46285", "accept": false, - "reason": "Position-id key warnings are not the same as CLIP text model structural changes." + "reason": "A position-id key issue across models is not the same as a CLIP text-model structure inconsistency." }, { - "left": "issue:45072", - "right": "issue:46832", + "left": "issue:43572", + "right": "issue:47436", "accept": false, - "reason": "bfloat16 dtype mismatch in inference is unrelated to RT-DETR feature-level indexing." + "reason": "Both mention missing config attributes, but they concern different models and different missing fields." }, { - "left": "issue:44464", - "right": "issue:45584", + "left": "issue:45072", + "right": "issue:46832", "accept": false, - "reason": "Chunked generation with compiled forward is a different path than Whisper empty-transcription failure after token alignment." + "reason": "A dtype mismatch in inference is unrelated to an RT-DETR feature-level indexing/type error." }, { - "left": "issue:43572", - "right": "issue:47436", + "left": "issue:43873", + "right": "issue:46050", "accept": false, - "reason": "Both are config attribute problems, but they affect different models and missing fields." + "reason": "Quantization/offloading behavior and save/load corruption of sinq-quantized models are distinct bugs." }, { - "left": "issue:43873", - "right": "issue:46050", + "left": "issue:44464", + "right": "issue:45584", "accept": false, - "reason": "Generic offloading/quantization behavior is not the same as sinq save/load corruption." + "reason": "Compiled-forward chunked generation inconsistency is unrelated to empty-transcription generation after token alignment." }, { "left": "issue:15354", "right": "issue:38175", "accept": false, - "reason": "TorchScript GeneratorExp export failure is unrelated to SigLIP2 zero-probability outputs." + "reason": "A torch.jit.script GeneratorExp export failure and unexpected zero probabilities in SigLIP2 are different issues." }, { "left": "issue:45245", "right": "issue:46710", "accept": false, - "reason": "Category-limit runtime error is unrelated to an accuracy regression after version upgrade." + "reason": "Category-count overflow and an accuracy regression after version upgrade are not the same bug." }, { "left": "issue:47013", "right": "issue:47254", "accept": false, - "reason": "xLSTM train-mode unpacking bug is distinct from PEFT + DeepSpeed checkpointing errors." + "reason": "An xLSTM train-mode unpacking crash and a PEFT/DeepSpeed checkpoint error are different code-path failures." }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "Qwen3.5 hidden_states bug and incorrect visual encoder save keys are different failure modes." + "reason": "Qwen3.5 output_hidden_states behavior and incorrect visual encoder keys on save_pretrained are unrelated regressions." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Template-node compilation error is unrelated to flash_attn import lookup on Python 3.13." - }, - { - "left": "issue:45229", - "right": "issue:45823", - "accept": false, - "reason": "CUDA OOM in multi-GPU inference is not the same as a PLE device mismatch in forward." + "reason": "A template-node compilation TypeError and a Python 3.13 flash_attn import KeyError are unrelated." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Gradient checkpointing ineffective with PEFT LoRA is distinct from checkpointing being disallowed in eval mode." + "reason": "Gradient checkpointing ineffective in one case and disallowed in eval mode in another are different behaviors, not duplicates." }, { - "left": "issue:28218", - "right": "issue:34567", + "left": "issue:45229", + "right": "issue:45823", "accept": false, - "reason": "Tokenizer spacing after added token and TrainerState token counters are unrelated." + "reason": "Multi-GPU CUDA OOM and device-mismatch with auto device mapping are different problems." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Continuous-batching multimodal crash with a string-to().to issue is different from KV-cache continuation tensor indexing." + "reason": "Continuous-batching multimodal crash and KV-cache continuation indexing bug are separate serving failures." }, { - "left": "issue:43844", - "right": "issue:44964", + "left": "issue:44849", + "right": "issue:46209", "accept": false, - "reason": "ZeRO-3 gradient explosion is unrelated to loading Phi-4 multimodal models." + "reason": "A Qwen3.5 hidden-states bug and an FP8 inference failure have no shared underlying cause." }, { - "left": "issue:44849", - "right": "issue:46209", + "left": "issue:43844", + "right": "issue:44964", "accept": false, - "reason": "Qwen3.5 hidden_states bug is unrelated to FP8 inference failure on B200." + "reason": "A DeepSpeed ZeRO-3 gradient anomaly and a model-loading failure for Phi-4 multimodal are unrelated." }, { - "left": "issue:45200", - "right": "issue:45229", + "left": "issue:28218", + "right": "issue:34567", "accept": false, - "reason": "Gemma4 mm_token_type_ids defaulting is different from multi-GPU OOM." + "reason": "Tokenizer spacing after added tokens and TrainerState token accounting are unrelated." }, { "left": "issue:45588", "right": "issue:46050", "accept": false, - "reason": "flash_attention.py sink-less model crash is unrelated to sinq quantized save/load breakage." + "reason": "A flash-attention crash on sink-less models and quantized save/load breakage are different bugs." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Missing mm_token_type_ids defaulting and multi-GPU OOM are unrelated issues." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "Missing internal flash-support import is not the same as meta-device loading breakage." + "reason": "A missing flash-attention import symbol and meta-device model loading failures are different regressions." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "SigLIP2 HF-vs-JAX discrepancy is unrelated to resume_from_checkpoint key mismatches." + "reason": "A SigLIP2 implementation discrepancy and a resume_from_checkpoint key mismatch do not describe the same bug." }, { "left": "issue:41720", "right": "issue:46710", "accept": false, - "reason": "Qwen3 auto device mapping CUDA assert is not an accuracy regression issue." + "reason": "Auto device mapping causing cudaErrorAssert and an accuracy regression are distinct failure modes." } ] }, @@ -9493,8 +9440,8 @@ "soft_pairs": [ "issue:44849|issue:47145", "issue:47332|issue:47404", - "issue:43937|issue:46701", "issue:43295|issue:43450", + "issue:43937|issue:46701", "issue:43065|issue:46369", "issue:44977|issue:46701", "issue:44554|issue:45446", @@ -9510,8 +9457,8 @@ "issue:43295|issue:44610", "issue:45823|issue:46192", "issue:45406|issue:46121", - "issue:44186|issue:46192", "issue:43450|issue:43575", + "issue:44186|issue:46192", "issue:37428|issue:43824", "issue:43550|issue:43844", "issue:45200|issue:45823", @@ -9521,218 +9468,215 @@ "issue:42915|issue:43824", "issue:43295|issue:43493" ], - "prepared_review_unit_hash": "d555ecd16058bb8e1ea1db0faf03d529280f6fc8c99ba234120d487f06565922", + "prepared_review_unit_hash": "87d0ce9f012e9f0a68350d54f686d5182b81a332a840bf6adc8b7dfd4f1f343c", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36413, - "estimated_input_tokens": 9104, - "estimated_eval_tokens": 18464 + "serialized_chars": 36408, + "estimated_input_tokens": 9102, + "estimated_eval_tokens": 18460 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a set of unrelated Transformers issues. Only one soft pair looks like a likely duplicate: the two missing-`pad_token_id` config regressions for different model families. The rest differ by model, code path, or symptom and should stay separate.", - "confidence": 0.89, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: the issues span unrelated bugs in model loading, attention backends, tokenizers/processors, offloading, compilation, and training/runtime behavior. The soft edges are mostly keyword overlap (e.g. FlashAttention, padding, offloading) rather than the same underlying defect, so they should be rejected.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue exists; these reports describe different bugs in different code paths and models, so there is no true duplicate center for the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good cluster canonical: the items span unrelated bugs across models, processors, offloading, attention, and import/config regressions.", + "best_issue_reason": "Issue 43873 is the strongest standalone representative only because it has the most discussion and inbound references, but it is still not a duplicate anchor for the rest of the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44849", "right": "issue:47145", "accept": false, - "reason": "Different bugs: Qwen3.5 hidden-states behavior vs FP8 frozen-checkpoint weight-scale loading." + "reason": "Different bugs: Qwen3.5 hidden-states behavior vs FP8 compressed-tensors checkpoints missing weight_scale application." }, { "left": "issue:47332", "right": "issue:47404", "accept": false, - "reason": "Both involve memory/offloading, but one is save-time disk offloading and the other is load-time device-map OOM during conversion." + "reason": "Both involve loading/offloading, but one is a save failure for disk-offloaded Qwen3-VL and the other is an OOM from conversion-op transients during device_map loading." }, { - "left": "issue:43937", - "right": "issue:46701", + "left": "issue:43295", + "right": "issue:43450", "accept": false, - "reason": "Unrelated models and failures: invalid GenerationConfig vs a diffusion pipeline output integration problem." + "reason": "Processor.tokenizer regression is unrelated to batched video shape handling." }, { - "left": "issue:43295", - "right": "issue:43450", + "left": "issue:43937", + "right": "issue:46701", "accept": false, - "reason": "Both are processor-related, but one is a tokenizer/processor API regression and the other is batched video shape handling." + "reason": "GLM-5 GenerationConfig validation and DiffusionGemma image-text pipeline failure are different code paths." }, { "left": "issue:43065", "right": "issue:46369", "accept": false, - "reason": "Same model family, but the symptoms differ: dummy Conv2d implementation vs memory overflow at model invocation." + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as a memory overflow in sam3-litetext." }, { "left": "issue:44977", "right": "issue:46701", "accept": false, - "reason": "Different model families and different failure modes; flash-attention generation issue is not the same as the diffusion pipeline bug." + "reason": "Flash-attention generation issue for Qwen3.5 is unrelated to the DiffusionGemma pipeline bug." }, { "left": "issue:44554", "right": "issue:45446", "accept": false, - "reason": "Different attention code paths: MPS correctness bug vs a PyTorch version check/import guard issue." + "reason": "MPS attention correctness and a PyTorch version-check import bug are separate problems." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "Different subsystems: bfloat16 dtype mismatches in inference vs Wav2Vec2 save/tokenization failure." + "reason": "Bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel are unrelated to Wav2Vec2 save/tokenization failures." }, { "left": "issue:45588", "right": "issue:45759", "accept": false, - "reason": "Both touch multimodal plumbing, but one crashes in flash-attention integration on `s_aux=None`; the other is a `from_config` text_config unwrap issue." + "reason": "A flash_attention helper crash on sink-less models is different from not unwrapping text_config in composite Qwen configs." }, { "left": "issue:43404", "right": "issue:46710", "accept": false, - "reason": "Different concrete problems: weight tying in Mistral3 vs an accuracy regression after a version upgrade." + "reason": "Untied lm_head weights in Mistral3ForConditionalGeneration is not the same as a DeepSeek accuracy regression after upgrade." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "One is a processor-tokenizer regression; the other is offloading behavior with quantization." + "reason": "Processor/tokenizer regression and quantization offloading issues are unrelated." }, { "left": "issue:44869", "right": "issue:46752", "accept": false, - "reason": "Different text-processing failures: Whisper timestamp decoding vs `apply_chat_template` on an empty list." + "reason": "Whisper timestamp decode crash and empty-chat-template IndexError are different text-processing bugs." }, { "left": "issue:43572", "right": "issue:46821", "accept": false, - "reason": "Both are missing-attribute errors, but they affect different configs and different missing fields (`pad_token_idx` vs `window_size`)." + "reason": "StableLm pad_token_idx regression and TimesFM window_size AttributeError are separate config issues." }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Completely unrelated: Flash Attention 2 test failures vs MLflow reporting after training job failure." + "reason": "Flash Attention 2 test failures do not match an MLFlow reporting issue after failed training jobs." }, { "left": "issue:28370", "right": "issue:34689", "accept": false, - "reason": "Both are loading-related, but one is a missing `vocab_file` on custom SentencePiece models and the other is a specific Llama 3.2 Vision loading regression." + "reason": "Missing vocab_file for custom SentencePiece models is unrelated to Llama 3.2 Vision loading breakage." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Different multimodal issues: video-input processing in Qwen3OmniMoe vs a Gemma 4 token-type-id defaulting problem." + "reason": "Qwen3OmniMoe video-processing ValueError and Gemma 4 mm_token_type_ids defaulting are different multimodal bugs." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Both involve processors, but the bugs are different: tokenizer access regression vs OmDet-Turbo producing the wrong image size." + "reason": "A processor.tokenizer regression is not the same as OmDet-Turbo producing the wrong image size." }, { "left": "issue:45823", "right": "issue:46192", "accept": false, - "reason": "Different bugs: Gemma4 device placement mismatch vs return_tensors being ignored when text kwargs are passed." + "reason": "Gemma4 PLE device mismatch and return_tensors being ignored are unrelated behaviors." }, { "left": "issue:45406", "right": "issue:46121", "accept": false, - "reason": "Unrelated failures: missing `_tokenizer` on Gemma4Processor vs rope-parameter validation type handling." + "reason": "Gemma4Processor missing _tokenizer and convert_rope_params_to_dict TypeError are different failures." }, { - "left": "issue:44186", - "right": "issue:46192", + "left": "issue:43450", + "right": "issue:43575", "accept": false, - "reason": "Both are tokenizer/input-preprocessing issues, but they target different APIs and failure modes." + "reason": "Batched video shape errors are unrelated to Qwen2-57B-A14B-Instruct tensor-parallel OOM." }, { - "left": "issue:43450", - "right": "issue:43575", + "left": "issue:44186", + "right": "issue:46192", "accept": false, - "reason": "Different areas: batched video processor shape bug vs tensor-parallel loading OOM." + "reason": "LayoutLMv2Tokenizer crashes on NER/padding are not the same as return_tensors being ignored with text_kwargs." }, { "left": "issue:37428", "right": "issue:43824", "accept": false, - "reason": "Both are ImportErrors, but they concern different symbols and different model families." + "reason": "These are both import errors, but for different missing symbols in different modules and model families." }, { "left": "issue:43550", "right": "issue:43844", "accept": false, - "reason": "Different training/runtime problems: torch.compile + SDPA failure vs gradient blow-up under ZeRO-3." + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to ZeRO-3 gradient growth." }, { "left": "issue:45200", "right": "issue:45823", "accept": false, - "reason": "Same model family, but separate bugs: missing text-only token-type defaults vs device-map forward mismatch." + "reason": "Gemma 4 token-type defaults and Gemma4 PLE device placement are separate problems." }, { "left": "issue:45520", "right": "issue:46178", "accept": false, - "reason": "Both are backend/import mapping issues, but they are for different backends and different error paths." + "reason": "flash_attn KeyError on Python 3.13 is unrelated to BACKENDS_MAPPING rejecting tensorflow_text." }, { "left": "issue:43334", "right": "issue:43525", - "accept": true, - "reason": "Strong match: both report the same missing `pad_token_id` attribute on newly introduced config classes, likely one shared regression in config plumbing." + "accept": false, + "reason": "Both mention pad_token_id, but they affect different config classes/models and are not the same concrete bug." }, { "left": "issue:27957", "right": "issue:28370", "accept": false, - "reason": "Both mention model loading/utilities, but XLM-R Flash Attention 2 and missing `vocab_file` are different bugs." + "reason": "XLMRoberta FlashAttention 2 support and missing vocab_file for custom SentencePiece models are unrelated." }, { "left": "issue:42915", "right": "issue:43824", "accept": false, - "reason": "Unrelated: FineGrainedFP8Config failure for Qwen3Moe vs a Qwen2.5-VL import error." + "reason": "FineGrainedFP8Config failure in Qwen3Moe is unrelated to the Qwen2_5_VL import error." }, { "left": "issue:43295", "right": "issue:43493", "accept": false, - "reason": "Different problems: processor/tokenizer regression vs SigLIP2 implementation discrepancy against JAX." + "reason": "Processor/tokenizer regression and SigLIP2 implementation discrepancy are distinct bugs." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "Grounded and conservative overall. The summary stays within the packet evidence, and the single accepted soft pair is the only one with a clearly shared failure mode (`pad_token_id` missing on config objects). The rejections are appropriately cautious given only titles/body snippets. " - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -9800,8 +9744,8 @@ ], "soft_pairs": [ "issue:42490|issue:43493", - "issue:39692|issue:46192", "issue:29870|issue:29942", + "issue:39692|issue:46192", "issue:43824|issue:43975", "issue:43278|issue:43856", "issue:43262|issue:43450", @@ -9811,205 +9755,205 @@ "issue:43653|issue:44112", "issue:43950|issue:46285", "issue:44737|issue:46894", - "issue:44442|issue:45198", "issue:43452|issue:46512", + "issue:44442|issue:45198", "issue:44479|issue:46192", "issue:43819|issue:44112", - "issue:44561|issue:46234", "issue:44466|issue:45127", "issue:43335|issue:43550", - "issue:45381|issue:46209", + "issue:44561|issue:46234", "issue:45200|issue:45325", "issue:43611|issue:47405", + "issue:45381|issue:46209", "issue:44898|issue:45412", "issue:45200|issue:46899", "issue:44610|issue:44857", "issue:37428|issue:43425" ], - "prepared_review_unit_hash": "98bd2ce2cf394c2c288c4029461e0a0c475744a2329548901306172440635aa2", + "prepared_review_unit_hash": "1608473db0e1fa76e9529b3391dab5486292ae23c3c54f6d81afbd72fff67edc", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 26, - "serialized_chars": 36008, - "estimated_input_tokens": 9002, - "estimated_eval_tokens": 18260 + "serialized_chars": 36014, + "estimated_input_tokens": 9004, + "estimated_eval_tokens": 18264 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab bag of unrelated Transformers bug reports rather than a duplicate set. None of the soft-pair candidates look like the same underlying bug or change, so I rejected all of them.", - "confidence": 0.94, - "canonical_issue_reason": "Issue #44521 is the best representative of the cluster only in a loose triage sense: it is active, has multiple inbound references, and sits in the broad chat-template/multimodal area that overlaps the only near-neighbor pair here.", + "summary": "The soft-pair suggestions are mostly false positives: the issues touch different models, APIs, or failure modes, and none of the PR-style merges are warranted here. I\u2019d use issue #44521 as the best representative issue because it is the most active and broadly referenced item in the set, but there is no true duplicate cluster.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #44521 is the strongest representative item by activity and inbound references, and it is an active, user-facing bug with broad discussion.", "canonical_pr_reason": null, - "best_issue_reason": "#44521 has the strongest overall triage signal among these heterogeneous issues (open, discussed, and referenced), making it the most plausible anchor issue even though the set is not truly duplicate-heavy.", + "best_issue_reason": "Issue #44521 is the most actionable and central issue in the set; it has the most discussion/ref activity and a clear bug report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42490", "right": "issue:43493", "accept": false, - "reason": "Different bugs: local-model path trailing-slash import warning vs SigLIP2 HF/JAX implementation discrepancy." + "reason": "Different problems: local-path loading warning vs. SigLIP2 implementation discrepancy." }, { - "left": "issue:39692", - "right": "issue:46192", + "left": "issue:29870", + "right": "issue:29942", "accept": false, - "reason": "One is a SigLIP2 docs/example bug; the other is a generic processor kwarg/return_tensors regression." + "reason": "Bart generate() decoder mask bug is unrelated to Flash Attention 2 test failures." }, { - "left": "issue:29870", - "right": "issue:29942", + "left": "issue:39692", + "right": "issue:46192", "accept": false, - "reason": "BART generate decoder_attention_mask UnboundLocalError is unrelated to Flash Attention 2 test failures." + "reason": "Documentation/example errors are unrelated to `return_tensors` being ignored when `text_kwargs` is passed." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "Qwen2.5-VL import error and DeepSeek detokenization regression affect different code paths and models." + "reason": "Qwen2.5-VL import error and DeepSeek detokenization are different code paths and symptoms." }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "Embedding dtype drift in evaluate is unrelated to Qwen3 MoE memory usage during training." + "reason": "Embedding dtype drift during evaluate is unrelated to Qwen3 MoE memory usage." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio chat-template sampling-rate default and batched video shape bug are different multimodal processor issues." + "reason": "Audio chat-template sampling-rate defaulting and batched video shape handling are separate processor bugs." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Detokenization bug is unrelated to Python 3.13 flash_attn import utilities failure." + "reason": "Detokenization corruption and Python 3.13 `flash_attn` import errors do not share the same underlying fix." }, { "left": "issue:28218", "right": "issue:35141", "accept": false, - "reason": "Tokenizer extra-space behavior after added tokens is not the same as output-embedding reinitialization on resize." + "reason": "Extra tokenizer space after added token is unrelated to output embedding reinitialization after resizing." }, { "left": "issue:44521", "right": "issue:45290", "accept": false, - "reason": "Both involve chat templates, but one is multimodal assistant-mask handling and the other is tool-call/no-content tokenization crash." + "reason": "Both involve `apply_chat_template`, but one is zeroed assistant masks for multimodal inputs and the other is a crash on tool-call messages; different bugs." }, { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "BigBird special-token registration and GraniteSpeech CI device-override flakiness are unrelated." + "reason": "BigBirdTokenizer special-token registration and a stale GraniteSpeech CI test are unrelated." }, { "left": "issue:43950", "right": "issue:46285", "accept": false, - "reason": "Non-persistent buffer corruption on from_pretrained is a different regression from CLIP text-model structural inconsistency." + "reason": "Non-persistent buffer corruption on `from_pretrained()` is unrelated to CLIP text model structure changes." }, { "left": "issue:44737", "right": "issue:46894", "accept": false, - "reason": "XLNet CPU arange/device placement bug is unrelated to CohereASR training-loss double-shift." + "reason": "XLNet CPU `arange` device bug and CohereASR training-loss double-shift are different model-specific issues." }, { - "left": "issue:44442", - "right": "issue:45198", + "left": "issue:43452", + "right": "issue:46512", "accept": false, - "reason": "FastSpeech2Conformer AutoTokenizer loading failure and Wav2Vec2 save_pretrained/tokenization failure are different model/tokenizer bugs." + "reason": "GGUF loading failures are unrelated to `from_config(dtype=...)` leaving `config.dtype` stale." }, { - "left": "issue:43452", - "right": "issue:46512", + "left": "issue:44442", + "right": "issue:45198", "accept": false, - "reason": "gguf_file/from_pretrained loading issue is not the same as from_config(dtype=...) leaving config.dtype stale." + "reason": "FastSpeech2Conformer AutoTokenizer loading failure and Wav2Vec2 save/tokenization failure are not the same bug." }, { "left": "issue:44479", "right": "issue:46192", "accept": false, - "reason": "Qwen video-input regression and return_tensors being ignored with text_kwargs are different API failures." + "reason": "Video input regression for Qwen models is unrelated to `return_tensors` handling in text kwargs." }, { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "DAC latent/forward mismatch and GraniteSpeech stale device override test are unrelated." - }, - { - "left": "issue:44561", - "right": "issue:46234", - "accept": false, - "reason": "Both mention import utilities, but they concern different missing helpers and different compatibility breaks." + "reason": "DAC latent reconstruction mismatch and a GraniteSpeech CI test failure are unrelated." }, { "left": "issue:44466", "right": "issue:45127", "accept": false, - "reason": "lm_head serialization inconsistency with device/tied weights is not the same as LoRA merge collapse with extended vocab." + "reason": "Serialization inconsistency for `lm_head.weight` is a different issue from LoRA merge collapse with extended vocabulary." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "SwitchTransformers sparse-layer config bug and Bamba torch.compile SDPA failure touch different code paths." + "reason": "SwitchTransformers sparse-layer config bug is unrelated to Bamba torch.compile/SDPA failure." }, { - "left": "issue:45381", - "right": "issue:46209", + "left": "issue:44561", + "right": "issue:46234", "accept": false, - "reason": "Qwen2.5-VL vision_position_ids regression and FP8 inference failure on B200 are unrelated issues." + "reason": "Removal of `is_torch_fx_available` breaking trust_remote_code models is unrelated to missing `is_falcon_mamba_ssm_available`." }, { "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Missing default mm_token_type_ids for text-only fine-tuning is different from Qwen2.5-VL rope/temporal position scaling." + "reason": "Gemma4 `mm_token_type_ids` defaults and Qwen2.5-VL rope-index scaling are different multimodal bugs." }, { "left": "issue:43611", "right": "issue:47405", "accept": false, - "reason": "Base_model_prefix loading break and silent model_type mismatch conversion skip are distinct loading/dispatch regressions." + "reason": "Base-model-prefix loading regression and silent wrong-Auto-class conversion are distinct loader failures." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "Qwen2.5-VL vision position IDs and FP8 inference on B200 are unrelated." }, { "left": "issue:44898", "right": "issue:45412", "accept": false, - "reason": "Perceiver interpolation failure at non-default resolution is unrelated to RT-DETR memory not being released." + "reason": "Perceiver non-default-resolution failure is unrelated to RT-DETR memory leak on deletion." }, { "left": "issue:45200", "right": "issue:46899", "accept": false, - "reason": "mm_token_type_ids defaulting and Gemma4 4-bit dtype-cast corruption are different multimodal processing bugs." + "reason": "Gemma4 text-only `mm_token_type_ids` defaulting and 4-bit quantization dtype-cast failures are different multimodal issues." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "OmDet-Turbo resolution mismatch and LwDetrImageLoss AMP crash are different failure modes." + "reason": "OmDet-Turbo processor size mismatch and LwDetrImageLoss AMP crash are separate problems." }, { "left": "issue:37428", "right": "issue:43425", "accept": false, - "reason": "Missing flash_supports_window_size import and Torch 2.10 incompatibility are separate environment/version issues." + "reason": "Missing flash-attention import symbol and Torch 2.10 incompatibility are not the same underlying issue." } ] }, @@ -10088,13 +10032,13 @@ "issue:43819|issue:45083", "issue:43335|issue:43653", "issue:43606|issue:45083", - "issue:41720|issue:46209", "issue:43525|issue:47436", "issue:44857|issue:45593", + "issue:41720|issue:46209", "issue:44792|issue:45563", "issue:28282|issue:33357", - "issue:46512|issue:46821", "issue:43335|issue:43504", + "issue:46512|issue:46821", "issue:43550|issue:44112", "issue:45923|issue:47405", "issue:38175|issue:45412", @@ -10105,210 +10049,216 @@ "issue:46431|issue:46566", "issue:44855|issue:45083", "issue:43335|issue:43606", - "issue:44617|issue:46697", "issue:43335|issue:44112", + "issue:44617|issue:46697", "issue:42490|issue:43854", "issue:43819|issue:44493", - "issue:46424|issue:46858", "issue:44292|issue:45520", + "issue:46424|issue:46858", "issue:43335|issue:45083" ], - "prepared_review_unit_hash": "14e7e60b1a7e3f8d7de859fbbdff1be99c931c015ee199081f99c8293060ca7a", + "prepared_review_unit_hash": "d6987566fca0ac4aca26197a72547c19912351c4f0ba51a1bf8b70d65a424242", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36667, - "estimated_input_tokens": 9167, - "estimated_eval_tokens": 18590 + "serialized_chars": 36672, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The pairwise candidates are mostly superficially similar but describe different bugs, models, or code paths. I would reject all soft edges and avoid collapsing this cluster into a duplicate set.", - "confidence": 0.88, - "canonical_issue_reason": "Issue #44843 is the clearest, most concrete bug report in the set: it names a specific failing code path, has an explicit offline-mode trigger, and is easier to use as a representative anchor than the others.", + "summary": "This cluster does not form a true duplicate set: the issues span unrelated models, docs, tokenizer, cache, device-mapping, and CI failures. All soft pairs should be rejected.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue stands out; the items are thematically unrelated and the soft links are only loose title similarity, not the same underlying bug.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #44843 is the strongest issue candidate overall because it is narrowly scoped, reproducible, and has a clear root-cause hypothesis; the rest are too heterogeneous to serve as better duplicates anchors.", + "best_issue_reason": "No global best issue is suitable as a representative duplicate target because there is no shared code-path or common fix across the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45200", "right": "issue:46932", "accept": false, - "reason": "Different bugs: Gemma text-only token-type defaults vs quantized vision/audio dtype casting." + "reason": "Different Gemma 4 problems: missing mm_token_type_ids default vs quantized vision/audio dtype casting." }, { "left": "issue:44843", "right": "issue:47427", "accept": false, - "reason": "One is offline hub lookup behavior; the other is initialization skipping on non-CUDA hardware." + "reason": "Offline Mistral regex patch bug vs redundant init skipping in _is_hf_initialized; unrelated code paths." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "Different model families and different failures: DAC latent reconstruction vs qwen3_omni_moe length helper." + "reason": "DAC latent decoding mismatch and qwen3_omni_moe length helper behavior are different bugs." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "Unrelated tokenizer special-token registration issue versus SwitchTransformers sparse-layer config bug." + "reason": "SwitchTransformers sparse-layer config bug and BigBirdTokenizer special-token registration are unrelated." }, { "left": "issue:43606", "right": "issue:45083", "accept": false, - "reason": "CPU offload device mismatch is unrelated to qwen3_omni_moe feature-length arithmetic." - }, - { - "left": "issue:41720", - "right": "issue:46209", - "accept": false, - "reason": "Both are hardware-related inference failures, but on different models, GPUs, and likely different code paths." + "reason": "CPU offload device mismatch in bark-small is unrelated to qwen3_omni_moe length calculation." }, { "left": "issue:43525", "right": "issue:47436", "accept": false, - "reason": "Both mention missing config fields, but they are different config classes and different absent attributes." + "reason": "Different missing-config-attribute bugs on different models: pad_token_id vs max_position_embeddings." }, { "left": "issue:44857", "right": "issue:45593", "accept": false, - "reason": "Different model components: float16 CUDA loss crash versus missing auxiliary losses when denoising is off." + "reason": "Different model losses: LwDetr float16 AMP crash vs D-FINE auxiliary-loss handling." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "Both are Qwen-related but the failures differ: cudaErrorAssert on auto device map vs FP8 inference on B200." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "Image-generation test failure and stale generate warning are distinct behaviors." + "reason": "janus image-generation test failure and paged-generate warning are not the same underlying issue." }, { "left": "issue:28282", "right": "issue:33357", "accept": false, - "reason": "PyTorch import failure is unrelated to a MacOS bus error loading a CLIP model." + "reason": "PyTorch missing ImportError and MacOS bus error in CLIP are entirely different failures." }, { - "left": "issue:46512", - "right": "issue:46821", + "left": "issue:43335", + "right": "issue:43504", "accept": false, - "reason": "Stale dtype metadata and TimesFM window_size AttributeError are separate config/constructor bugs." + "reason": "SwitchTransformers sparse-layer creation and Beit legacy-field loading are unrelated model-specific bugs." }, { - "left": "issue:43335", - "right": "issue:43504", + "left": "issue:46512", + "right": "issue:46821", "accept": false, - "reason": "Different models and failures: sparse encoder-layer creation versus BEiT legacy field loading." + "reason": "Stale config.dtype after from_config is unrelated to TimesFM window_size attribute errors." }, { "left": "issue:43550", "right": "issue:44112", "accept": false, - "reason": "torch.compile/SDPA failure is not the same as a stale GraniteSpeech CI test." + "reason": "Bamba torch.compile/SDPA failure and GraniteSpeech stale device-override CI test are different problems." }, { "left": "issue:45923", "right": "issue:47405", "accept": false, - "reason": "Missing gradient-checkpointing flag is unrelated to silent wrong-auto-class checkpoint loading." + "reason": "Missing supports_gradient_checkpointing flag is unrelated to silent conversion skipping on model_type mismatch." }, { "left": "issue:38175", "right": "issue:45412", "accept": false, - "reason": "Unexpected probabilities and RT-DETR memory leaks are unrelated issues." + "reason": "Zero probabilities in SigLIP2 and RT-DETR memory retention have no shared fix path." }, { "left": "issue:45759", "right": "issue:46285", "accept": false, - "reason": "Multimodal text_config unwrapping and CLIP structure inconsistency are different API regressions." + "reason": "Composite Qwen from_config unwrapping and CLIP text_model structure drift are different model-API issues." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs regressions, but they affect different pages and different deprecated pipeline behaviors." + "reason": "Both are docs issues, but they cover different APIs and different deprecated behaviors." }, { "left": "issue:29942", "right": "issue:32090", "accept": false, - "reason": "Flash Attention 2 test failures and _gpu_broadcast_one TypeError are unrelated." + "reason": "Flash Attention 2 test failures and _gpu_broadcast_one NoneType errors are unrelated trainer/runtime bugs." }, { "left": "issue:44206", "right": "issue:45584", "accept": false, - "reason": "Unsupported processor arg and empty-transcription generation failure are different code paths." + "reason": "Unsupported center arg in feature extractor and Whisper empty-transcription failure are different code paths." }, { "left": "issue:46431", "right": "issue:46566", "accept": false, - "reason": "Missing chat template on a processor is unrelated to weights remaining on meta with auto device mapping." + "reason": "Missing chat template error and DiffusionGemma meta-device loading are unrelated loading issues." }, { "left": "issue:44855", "right": "issue:45083", "accept": false, - "reason": "Python 3.13 import parsing failure is not the same as qwen3_omni_moe feature-length logic." + "reason": "DebertaV2 import IndentationError and qwen3_omni_moe helper behavior are unrelated." }, { "left": "issue:43335", "right": "issue:43606", "accept": false, - "reason": "SwitchTransformers sparse-layer config and Bark CPU offload mismatch are unrelated." + "reason": "SwitchTransformers sparse-layer config and bark-small CPU offload device mismatch do not share a fix." + }, + { + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug and GraniteSpeech stale device override test are unrelated." }, { "left": "issue:44617", "right": "issue:46697", "accept": false, - "reason": "CUDA OOM in Sam3Video is not the same bug as Swin2SR over-padding." + "reason": "Sam3Video OOM and Swin2SR over-padding are unrelated issues." }, { "left": "issue:42490", "right": "issue:43854", "accept": false, - "reason": "Trailing-slash import warning and GLM-4.7-Flash unit-test loading failure are different issues." + "reason": "Trailing-slash import warning and GLM-4.7-Flash unit-test load failure are different loading problems." }, { "left": "issue:43819", "right": "issue:44493", "accept": false, - "reason": "DAC latent mismatch and unexpected position-id key warnings do not share a concrete code path." + "reason": "DAC.from_latents mismatch and unexpected position-id key warnings are distinct model bugs." }, { - "left": "issue:46424", - "right": "issue:46858", + "left": "issue:44292", + "right": "issue:45520", "accept": false, - "reason": "Cache max length handling and static-cache slice arithmetic on non-CUDA backends are distinct cache bugs." + "reason": "Qwen-3-8B-NVFP4 runtime failure and flash_attn Python 3.13 import KeyError are unrelated." }, { - "left": "issue:44292", - "right": "issue:45520", + "left": "issue:46424", + "right": "issue:46858", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime failure and flash_attn import KeyError are unrelated." + "reason": "Cache max_cache_length behavior and device-tensor seq_length arithmetic are different cache bugs." }, { "left": "issue:43335", "right": "issue:45083", "accept": false, - "reason": "Different models and mechanics: sparse-layer creation versus helper-length computation." + "reason": "SwitchTransformers sparse-layer creation and qwen3_omni_moe feature-length helper behavior are unrelated." } ] }, @@ -10335,11 +10285,9 @@ "issue:10000", "issue:15354", "issue:25251", - "issue:26413", "issue:27957", "issue:29127", "issue:29942", - "issue:31501", "issue:32090", "issue:33357", "issue:34689", @@ -10373,11 +10321,13 @@ "issue:45083", "issue:45290", "issue:45310", + "issue:45459", "issue:45478", "issue:45588", "issue:45593", "issue:45823", "issue:45859", + "issue:46234", "issue:46326", "issue:47000" ], @@ -10389,244 +10339,244 @@ "issue:44206|issue:45823", "issue:39692|issue:44493", "issue:44448|issue:44743", - "issue:39692|issue:43526", "issue:10000|issue:15354", - "issue:34689|issue:35141", + "issue:39692|issue:43526", "issue:45290|issue:46326", + "issue:34689|issue:35141", "issue:44514|issue:44521", "issue:43720|issue:44112", "issue:43701|issue:45083", "issue:44857|issue:45072", "issue:29127|issue:32090", "issue:44448|issue:44779", - "issue:43550|issue:45083", "issue:32090|issue:33357", + "issue:43550|issue:45083", "issue:38175|issue:44493", "issue:42490|issue:43575", - "issue:43278|issue:43825", "issue:37428|issue:42175", - "issue:43278|issue:43329", - "issue:25251|issue:32090", + "issue:43278|issue:43825", "issue:27957|issue:34689", + "issue:43278|issue:43329", "issue:45310|issue:45478", + "issue:25251|issue:32090", "issue:29942|issue:35141", "issue:45593|issue:47000", - "issue:26413|issue:31501" + "issue:45459|issue:46234" ], - "prepared_review_unit_hash": "dc6c6b7de5bd55e981950942a644e9d761208add701b82d4da7416ec4b971c9d", + "prepared_review_unit_hash": "3b0b717e9bfac1fbe8181af91fc915c66c36bfbef3c1ede7ddb04fdd380bb30a", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36816, - "estimated_input_tokens": 9204, - "estimated_eval_tokens": 18664 + "serialized_chars": 36847, + "estimated_input_tokens": 9212, + "estimated_eval_tokens": 18680 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are mostly unrelated single-issue reports across different models/code paths. The only clear duplicate soft edge is the pair of identical Qwen3.5 Moe `from_pretrained` regression reports; all other soft pairs should stay separate.", - "confidence": 0.96, - "canonical_issue_reason": "issue:45310 is the earliest clear report of the Qwen3.5 Moe `from_pretrained` regression, and issue:45478 is an almost exact duplicate with the same title and failure mode.", + "summary": "Most candidate edges are not true duplicates: they share a subsystem or symptom category, but the underlying failures differ. The only clear duplicate is the repeated Qwen3.5 Moe `from_pretrained` regression report (issues 45310 and 45478).", + "confidence": 0.88, + "canonical_issue_reason": "Issue 45310 is the better canonical issue for the accepted duplicate pair because it was filed earlier, has more context, and already has inbound reference activity.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the strongest issue representative because it has the same concrete regression as issue:45478 and appears to be the original report.", + "best_issue_reason": "Issue 45310 is the best issue representative overall among the accepted duplicate edge because it names the concrete regression clearly and appears to be the original report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43388", "right": "issue:43526", "accept": false, - "reason": "Both mention label handling, but they hit different functions and different bugs (`gather_for_metrics` truncation vs `reduce_labels` returning one label)." + "reason": "Both involve label handling, but one is `gather_for_metrics` dropping tuple labels and the other is `reduce_labels` in `BeitImageProcessorFast`; different code paths and bugs." }, { "left": "issue:45588", "right": "issue:45859", "accept": false, - "reason": "Different failure sites and code paths: flash-attention integration on `s_aux=None` vs Qwen3.5 Moe rotary embedding CPU-offload incompatibility." + "reason": "Both touch flash-attention/CPU-offload areas, but one is an `AttributeError` in `integrations/flash_attention.py` for sink-less models and the other is a rotary embedding incompatibility in Qwen3.5 MoE." }, { "left": "issue:37428", "right": "issue:44351", "accept": false, - "reason": "These are separate import errors for different symbols (`_flash_supports_window_size` vs `HybridCache`) and do not describe the same bug." + "reason": "Both are import-related, but they concern different missing symbols (`_flash_supports_window_size` vs `HybridCache`) and different failure modes." }, { "left": "issue:43606", "right": "issue:44857", "accept": false, - "reason": "Completely different models and symptoms: CPU-offload device mismatch in Bark vs AMP/CUDA crash in LwDetrImageLoss." + "reason": "These are unrelated runtime crashes: one is a device mismatch with CPU offload for `suno/bark-small`, the other is an AMP/CUDA loss crash in `LwDetrImageLoss`." }, { "left": "issue:44206", "right": "issue:45823", "accept": false, - "reason": "One is a bad argument in `LasrFeatureExtractor`, the other is a device-mapping mismatch in Gemma4 PLE forward; not the same issue." + "reason": "Both mention mismatches, but one is a `LasrFeatureExtractor` center-argument regression and the other is a `device_map=\"auto\"` forward device mismatch in Gemma4 PLE." }, { "left": "issue:39692", "right": "issue:44493", "accept": false, - "reason": "Docs example/model-processor mismatch and quantization failure is unrelated to unexpected `position_id` key behavior across many models." + "reason": "One is a SigLIP2 docs/example problem, the other is a broad position-id key warning across many models; not the same concrete bug." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Both involve model behavior changes, but the concrete bugs are different: Pegasus output drift vs Qwen recurrent state reset with cache." - }, - { - "left": "issue:39692", - "right": "issue:43526", - "accept": false, - "reason": "A SigLIP2 documentation-example problem is unrelated to `BeitImageProcessorFast.reduce_labels` returning only one label." + "reason": "A Pegasus output-difference regression and a Qwen3 modular recurrent-state reset bug are different model behaviors and code paths." }, { "left": "issue:10000", "right": "issue:15354", "accept": false, - "reason": "Different models, different stacks, and different failures: German DistilBertModel issue vs torch.jit GeneratorExp export error for ViT." + "reason": "Both are model/export issues, but one is a German DistilBERT issue and the other is a TorchScript `GeneratorExp` export problem for ViT." }, { - "left": "issue:34689", - "right": "issue:35141", + "left": "issue:39692", + "right": "issue:43526", "accept": false, - "reason": "Llama 3.2 Vision load regression and token-embedding reinitialization are different bugs with different triggers." + "reason": "The shared theme is image-processing label handling, but the SigLIP2 example errors and BeitImageProcessorFast `reduce_labels` bug are not the same defect." }, { "left": "issue:45290", "right": "issue:46326", "accept": false, - "reason": "Both touch chat templating, but one crashes on tool-call assistant messages and the other concerns `continue_final_message` for mistral_common tokenizers; not the same concrete bug." + "reason": "Both concern chat templating, but one crashes on tool-call messages with no content while the other is about `continue_final_message` for mistral_common tokenizers." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "These are separate model-loading/embedding issues: Llama 3.2 Vision load failure versus token-embedding resize reinitialization when `tie_word_embedding=False`." }, { "left": "issue:44514", "right": "issue:44521", "accept": false, - "reason": "Both are `apply_chat_template` issues, but one is a batching/padding crash and the other is incorrect all-zero assistant masks for multimodal inputs." + "reason": "Both are `apply_chat_template` bugs, but one crashes on batched input with `padding=False` and the other returns zero assistant masks for multimodal inputs." }, { "left": "issue:43720", "right": "issue:44112", "accept": false, - "reason": "Different areas entirely: BitNet packed-weight loading vs a stale GraniteSpeech CI test failure." + "reason": "Both are CI/loading regressions, but one is about unpacking BitNet weights during accelerate loading and the other is a stale GraniteSpeech device override test." }, { "left": "issue:43701", "right": "issue:45083", "accept": false, - "reason": "Resume-from-checkpoint key mismatch is unrelated to `_get_feat_extract_output_lengths` behavior in qwen3_omni_moe." + "reason": "A `resume_from_checkpoint` key mismatch is unrelated to the Qwen3 Omni MoE feature-extraction length helper behaving unexpectedly." }, { "left": "issue:44857", "right": "issue:45072", "accept": false, - "reason": "Both are dtype-related, but they affect different models/code paths and different precisions (`float16` AMP crash vs `bfloat16` inference mismatch)." + "reason": "Both involve numeric precision issues, but the failing components and scenarios differ: `LwDetrImageLoss` under float16 AMP versus SwitchTransformers/TimmWrapperModel bfloat16 inference mismatches." }, { "left": "issue:29127", "right": "issue:32090", "accept": false, - "reason": "One is a clearer `layoutlmv3` error-message issue; the other is a Trainer broadcast `NoneType` TypeError. Different bugs." + "reason": "Different problems: one is a LayoutLMv3 box-validation error message, the other is a Trainer `_gpu_broadcast_one` NoneType issue." }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Different model families and symptoms: Pegasus output changes vs Deepseek tokenizer regression in v5." + "reason": "Different regressions in different subsystems: Pegasus output changes versus DeepSeek tokenizer behavior changes." }, { - "left": "issue:43550", - "right": "issue:45083", + "left": "issue:32090", + "right": "issue:33357", "accept": false, - "reason": "`torch.compile` + SDPA failure in Bamba is unrelated to a feature-extractor length helper bug in qwen3_omni_moe." + "reason": "Both are runtime failures, but the causes differ completely: Trainer broadcast of `NoneType` versus MacOS bus error with a CLIP community model." }, { - "left": "issue:32090", - "right": "issue:33357", + "left": "issue:43550", + "right": "issue:45083", "accept": false, - "reason": "A Trainer `_gpu_broadcast_one` TypeError and a MacOS bus error with a community CLIP model are not the same failure." + "reason": "These are distinct compile/runtime bugs: Bamba SDPA with `torch.compile` versus Qwen3 Omni MoE feature-length calculation." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 zero probabilities and unexpected `position_id` keys are distinct symptoms with no clear shared code path." + "reason": "Unexpected zero probabilities in SigLIP2 and unexpected position-id keys are different symptoms with no clear shared code-path." }, { "left": "issue:42490", "right": "issue:43575", "accept": false, - "reason": "Local-path import warning is unrelated to Qwen2-57B-A14B-Instruct tensor-parallel OOM." - }, - { - "left": "issue:43278", - "right": "issue:43825", - "accept": false, - "reason": "Embedding dtype drift during eval and a pipeline translation-deprecation warning are unrelated." + "reason": "A local-path trailing-slash import warning and a tensor-parallel OOM when loading Qwen2-57B-A14B-Instruct are unrelated." }, { "left": "issue:37428", "right": "issue:42175", "accept": false, - "reason": "A missing flash-attention helper import is unrelated to TensorFlow not being installed in the `.[torch]` extra." + "reason": "One is a missing internal flash-attention helper import, the other is packaging/back-end dependency coverage for TensorFlow under `pip install .[torch]`." }, { "left": "issue:43278", - "right": "issue:43329", + "right": "issue:43825", "accept": false, - "reason": "One concerns dtype changes in embeddings; the other is an undefined-variable bug in multimodal token counting. Different code paths." + "reason": "An embedding dtype drift during evaluate and a pipeline error message incorrectly implying translation support are different issues." }, { - "left": "issue:25251", - "right": "issue:32090", + "left": "issue:27957", + "right": "issue:34689", "accept": false, - "reason": "Pipeline `top_k` nesting behavior and Trainer broadcast errors are unrelated." + "reason": "Both involve model support changes, but XLMRoberta Flash Attention 2 and Llama 3.2 Vision loading failures are different problems." }, { - "left": "issue:27957", - "right": "issue:34689", + "left": "issue:43278", + "right": "issue:43329", "accept": false, - "reason": "Both are model/version regressions, but they affect different models and failure modes; no evidence they are the same bug." + "reason": "A dtype-change regression and a multimodal token-count helper using undefined video variables are unrelated bugs." }, { "left": "issue:45310", "right": "issue:45478", "accept": true, - "reason": "Same title, same model family, same `transformers>=5.4.0` / Qwen3.5 Moe `from_pretrained` regression; clear duplicate." + "reason": "These reports have the same title, same model family, and the same concrete `from_pretrained` regression for transformers>=5.4.0." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline `top_k` nesting behavior and Trainer GPU broadcast of `NoneType` are different features and failure modes." }, { "left": "issue:29942", "right": "issue:35141", "accept": false, - "reason": "Flash Attention 2 test failures and token-embedding reinitialization are separate issues." + "reason": "Failing Flash Attention 2 tests and token-embedding reinitialization on resize are unrelated; one is test/FA2 related and the other is model initialization logic." }, { "left": "issue:45593", "right": "issue:47000", "accept": false, - "reason": "D-FINE auxiliary-loss behavior and RTDetrLoss infeasible cost-matrix crashes are different detection-training bugs." + "reason": "Both are detector training losses, but one is D-FINE losing auxiliary losses when denoising is off and the other is RTDetrLoss infeasible cost matrix." }, { - "left": "issue:26413", - "right": "issue:31501", + "left": "issue:45459", + "right": "issue:46234", "accept": false, - "reason": "Checkpoint-resume epoch-iterator failure and FineWeb training not starting are not the same concrete bug." + "reason": "Tokenizer protobuf error handling and a missing `is_falcon_mamba_ssm_available` utility export are unrelated import/exception issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet. The duplicate call on issue:45310/45478 is conservative and supported by identical titles plus the earlier creation date. The other soft-edge rejects stay cautious and do not overclaim sameness from superficial similarity alone." + "feedback": "The analysis is conservative and grounded in the packet. The only accepted edge, issue:45310 \u2194 issue:45478, is strongly supported by identical titles and matching Qwen3.5 MoE from_pretrained regression wording. The rejected pairs are generally framed cautiously as related symptoms rather than duplicates, which fits the evidence available." }, "evaluator_used": true, "retried": false, @@ -10647,10 +10597,12 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ + "issue:26413", "issue:29870", "issue:29925", "issue:29942", "issue:30064", + "issue:31501", "issue:32090", "issue:33290", "issue:33357", @@ -10663,6 +10615,7 @@ "issue:43299", "issue:43381", "issue:43526", + "issue:43701", "issue:43873", "issue:43927", "issue:44186", @@ -10675,7 +10628,6 @@ "issue:44589", "issue:44779", "issue:44855", - "issue:44871", "issue:44977", "issue:45127", "issue:45198", @@ -10683,13 +10635,11 @@ "issue:45245", "issue:45290", "issue:45381", - "issue:45459", "issue:45588", "issue:45698", "issue:45823", "issue:45920", "issue:46192", - "issue:46234", "issue:46399", "issue:46424", "issue:46655", @@ -10697,238 +10647,238 @@ "issue:47436" ], "soft_pairs": [ - "issue:45459|issue:46234", - "issue:45127|issue:45245", + "issue:26413|issue:31501", "issue:29870|issue:32090", + "issue:45127|issue:45245", "issue:44360|issue:44561", - "issue:44448|issue:44977", "issue:45290|issue:45381", - "issue:40444|issue:43526", "issue:32090|issue:33290", + "issue:44448|issue:44977", + "issue:40444|issue:43526", "issue:29870|issue:33290", - "issue:44464|issue:46424", "issue:44589|issue:47436", + "issue:44464|issue:46424", "issue:44530|issue:47276", "issue:43278|issue:43381", "issue:44855|issue:46399", "issue:29925|issue:34689", "issue:42490|issue:43873", - "issue:44977|issue:45229", - "issue:45920|issue:46192", "issue:43927|issue:45698", + "issue:44977|issue:45229", "issue:43257|issue:43299", + "issue:45920|issue:46192", "issue:45290|issue:46655", "issue:29942|issue:30064", "issue:33357|issue:39692", - "issue:44186|issue:45198", "issue:44561|issue:44779", - "issue:43873|issue:45229", + "issue:44186|issue:45198", "issue:45588|issue:45823", - "issue:44871|issue:44977", - "issue:44448|issue:44466" + "issue:43873|issue:45229", + "issue:44448|issue:44466", + "issue:43381|issue:43701" ], - "prepared_review_unit_hash": "ac0a0cfea80a7cd727242a53ef0df49e5ec6b0247630a606949c72f36b585404", + "prepared_review_unit_hash": "08aec35d81c8e66774411e377124646c74de57012bbe67d430ec899069cfdbb7", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36547, - "estimated_input_tokens": 9137, - "estimated_eval_tokens": 18530 + "serialized_chars": 36438, + "estimated_input_tokens": 9110, + "estimated_eval_tokens": 18476 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The candidate pairs are mostly unrelated issue reports across different models, subsystems, and failure modes. I would not merge any of the soft-edge pairs; if a single issue must represent the set, the broadest/high-signal regression is the AutoTokenizer v5 regression report.", - "confidence": 0.88, - "canonical_issue_reason": "issue:45920 is the strongest representative bug report here: it describes a broad v5 regression affecting multiple models, has multiple inbound references, and a clear concrete symptom (wrong token IDs).", + "summary": "This cluster is heterogeneous: the issues span unrelated training, generation, tokenizer, model-loading, quantization, and documentation problems. None of the soft edge candidates look like the same underlying bug or change, so no duplicates are accepted. If a single issue must represent the cluster, the broadest and most central is the v5 tokenizer regression in issue 45920.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 45920 is the broadest, most cross-cutting problem in the set: it describes a v5 regression affecting AutoTokenizer across multiple model families, which makes it the best representative of the cluster\u2019s dominant compatibility/regression theme.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 is the best issue to keep as a canonical anchor because it is broad, well-scoped, and impacts multiple downstream users/models.", + "best_issue_reason": "Issue 45920 is the most suitable global issue because it is the most general and impactful regression among the listed issues, with multiple affected model families and clear evidence of broader fallout.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45459", - "right": "issue:46234", + "left": "issue:26413", + "right": "issue:31501", "accept": false, - "reason": "Tokenizer protobuf decode-error masking and a missing import_utils symbol are different failure classes with no shared code path." + "reason": "Both are training-start failures, but the specific symptoms differ: resume-from-checkpoint epoch_iterator failure versus FineWeb SLM startup not beginning. No shared code-path is evident." }, { - "left": "issue:45127", - "right": "issue:45245", + "left": "issue:29870", + "right": "issue:32090", "accept": false, - "reason": "LoRA merge / extended-vocab collapse is unrelated to the 2^24 categories limit error." + "reason": "Different failure modes in different parts of the stack: BART generate decoder_attention_mask assignment bug versus Trainer GPU broadcast NoneType error." }, { - "left": "issue:29870", - "right": "issue:32090", + "left": "issue:45127", + "right": "issue:45245", "accept": false, - "reason": "BART generate() UnboundLocalError and Trainer _gpu_broadcast_one NoneType handling are separate bugs in different paths." + "reason": "One is a LoRA merge / tied-weights collapse issue, the other is a category-cardinality runtime limit. They are unrelated." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "DSA indexer ReLU discussion and trust_remote_code breakage from removing is_torch_fx_available are unrelated." + "reason": "A DSA indexer/ReLU discussion and a v5 utility-removal compatibility break are not the same bug or fix." }, { - "left": "issue:44448", - "right": "issue:44977", + "left": "issue:45290", + "right": "issue:45381", "accept": false, - "reason": "Pegasus output drift and Qwen3.5 flash-attention generation failure are different model/regression reports." + "reason": "Chat template tool-call handling and Qwen2.5-VL video position IDs are distinct multimodal bugs with different code paths." }, { - "left": "issue:45290", - "right": "issue:45381", + "left": "issue:32090", + "right": "issue:33290", "accept": false, - "reason": "Chat-template tool-call crashes and Qwen2.5-VL video vision_position_ids issues do not share the same code path." + "reason": "Unsupported NoneType GPU broadcast and Adafactor/DeepSpeed OOM are separate Trainer/optimizer issues." }, { - "left": "issue:40444", - "right": "issue:43526", + "left": "issue:44448", + "right": "issue:44977", "accept": false, - "reason": "Multi-image iterable finetuning failure and BeitImageProcessorFast reduce_labels behavior are unrelated." + "reason": "Both are regression reports, but they target different models and different subsystems: Pegasus output drift versus Qwen3.5 flash-attention generation." }, { - "left": "issue:32090", - "right": "issue:33290", + "left": "issue:40444", + "right": "issue:43526", "accept": false, - "reason": "A NoneType broadcast crash in Trainer is not the same issue as an Adafactor+deepspeed OOM." + "reason": "Multiple-images-per-prompt IterableDataset failure and BeitImageProcessorFast label reduction are unrelated image-processing bugs." }, { "left": "issue:29870", "right": "issue:33290", "accept": false, - "reason": "The BART generate() assignment bug is unrelated to the deepspeed optimizer memory blowup." + "reason": "A generate-time UnboundLocalError and a DeepSpeed OOM are clearly different problems." }, { - "left": "issue:44464", - "right": "issue:46424", + "left": "issue:44589", + "right": "issue:47436", "accept": false, - "reason": "Chunked generation inconsistencies with compiled forward and cache max length handling are separate cache/generation issues." + "reason": "Float8 storage lookup failure and missing max_position_embeddings on Qwen2_5OmniConfig are unrelated runtime/config errors." }, { - "left": "issue:44589", - "right": "issue:47436", + "left": "issue:44464", + "right": "issue:46424", "accept": false, - "reason": "Float8 storage lookup failure and Qwen2.5OmniConfig missing max_position_embeddings are unrelated load-time errors." + "reason": "Compiled-forward chunked generation inconsistency and cache max length being honored only for quantized cache are separate cache/generation behaviors." }, { "left": "issue:44530", "right": "issue:47276", "accept": false, - "reason": "PagedAttentionCache invalid group type and a misleading synced_gpus warning are different batching/caching problems." + "reason": "A PagedAttentionCache crash and a misleading warning message are not the same underlying bug." }, { "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "Embedding dtype drift in eval and gradient checkpointing disallowed in eval mode are not the same bug." + "reason": "Embedding dtype drift during evaluation and gradient checkpointing being disallowed in eval mode are different eval-time issues." }, { "left": "issue:44855", "right": "issue:46399", "accept": false, - "reason": "Python 3.13 JIT parsing/IndentationError and a missing Qwen3.5 kernel decorator are unrelated." + "reason": "Both mention decorator-related failures, but they involve different models and different causes: Python 3.13 parsing versus missing decorator registration for rotary embeddings." }, { "left": "issue:29925", "right": "issue:34689", "accept": false, - "reason": "SigLIP tokenizer fast-mode enforcement and Llama 3.2 vision model loading breakage are separate problems." + "reason": "SigLIP tokenizer fast-mode enforcement and Llama 3.2 Vision model-loading regression are not the same bug." }, { "left": "issue:42490", "right": "issue:43873", "accept": false, - "reason": "Trailing-slash local model import warnings and quantization offloading failures do not match." + "reason": "Trailing-slash import warning and quantization offloading problems are unrelated." }, { - "left": "issue:44977", - "right": "issue:45229", + "left": "issue:43927", + "right": "issue:45698", "accept": false, - "reason": "Qwen3.5 flash-attention generation failure and Gemma4 multi-GPU CUDA OOM are different issues." + "reason": "Both involve save/load behavior, but one is config token IDs being lost while the other is loading the wrong custom module. Different objects and failure modes." }, { - "left": "issue:45920", - "right": "issue:46192", + "left": "issue:44977", + "right": "issue:45229", "accept": false, - "reason": "Wrong token IDs in AutoTokenizer and return_tensors being ignored are distinct tokenizer API regressions." + "reason": "Qwen3.5 flash-attention generation issues and Gemma4 multi-GPU CUDA OOM are separate model/runtime problems." }, { - "left": "issue:43927", - "right": "issue:45698", + "left": "issue:43257", + "right": "issue:43299", "accept": false, - "reason": "DiaConfig losing custom token IDs and loading the wrong custom module after save_pretrained are different serialization bugs." + "reason": "Both concern Qwen3 MOE loading, but one is accelerate+deepspeed weight conversion and the other is a v5 loading break for Qwen3VL MOE models. Too different to merge." }, { - "left": "issue:43257", - "right": "issue:43299", + "left": "issue:45920", + "right": "issue:46192", "accept": false, - "reason": "Qwen3 MOE weight conversion with accelerate+deepspeed and Qwen3VL MOE loading in v5 are related thematically but not the same concrete failure." + "reason": "Wrong token IDs across multiple models is a tokenizer regression; silently ignored return_tensors with text_kwargs is a separate API-argument bug." }, { "left": "issue:45290", "right": "issue:46655", "accept": false, - "reason": "Chat template tool-call handling and video processor padding calculation are unrelated." + "reason": "Chat template tool-call crashes and SmolVLM video padding calculation are unrelated multimodal bugs." }, { "left": "issue:29942", "right": "issue:30064", "accept": false, - "reason": "Flash Attention 2 test failures and void segmentation map processing are different test/processor bugs." + "reason": "Flash Attention 2 test failures and void segmentation-map processing are unrelated test/image-processing issues." }, { "left": "issue:33357", "right": "issue:39692", "accept": false, - "reason": "MacOS bus error with a CLIP model and SigLIP2 docs example errors are unrelated." - }, - { - "left": "issue:44186", - "right": "issue:45198", - "accept": false, - "reason": "LayoutLMv2Tokenizer NER/padding crashes and Wav2Vec2 save/tokenization failures are distinct tokenizer/model issues." + "reason": "A MacOS bus error with a CLIP model and documentation example mistakes for SigLIP2 are not the same underlying issue." }, { "left": "issue:44561", "right": "issue:44779", "accept": false, - "reason": "Removal of is_torch_fx_available breaking trust_remote_code and DeepSeek tokenizer regression are unrelated v5 issues." + "reason": "Both are v5 regressions, but one is a missing utility breaking trust_remote_code and the other is a tokenizer correctness regression. Different fixes." }, { - "left": "issue:43873", - "right": "issue:45229", + "left": "issue:44186", + "right": "issue:45198", "accept": false, - "reason": "Quantization offloading problems and Gemma4 multi-GPU OOM are different resource-management failures." + "reason": "LayoutLMv2 tokenizer crashes and Wav2Vec2 save_pretrained/tokenization failures involve different model families and code paths." }, { "left": "issue:45588", "right": "issue:45823", "accept": false, - "reason": "flash_attention.py s_aux=None crash and Gemma4 device-map mismatch are separate forward-path bugs." + "reason": "A flash_attention.py AttributeError on sink-less models and a Gemma4 device_map mismatch during forward are distinct runtime issues." }, { - "left": "issue:44871", - "right": "issue:44977", + "left": "issue:43873", + "right": "issue:45229", "accept": false, - "reason": "Gemma-3 eos_token_id inconsistency and Qwen3.5 flash-attention generation failure are unrelated." + "reason": "Quantization offloading not working and multi-GPU CUDA OOM are related only at a very broad memory level, not the same bug." }, { "left": "issue:44448", "right": "issue:44466", "accept": false, - "reason": "Pegasus v4/v5 output differences and serialized lm_head.weight inconsistencies are different regression reports." + "reason": "Pegasus output differences between versions and lm_head.weight serialization differences depending on device are separate v5 regression reports." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are different checkpointing/training issues." } ] }, @@ -10984,6 +10934,7 @@ "issue:44479", "issue:44568", "issue:44743", + "issue:44871", "issue:44977", "issue:45072", "issue:45198", @@ -10997,264 +10948,256 @@ "issue:45803", "issue:46121", "issue:46421", - "issue:46693", "issue:46858", "issue:47276" ], "soft_pairs": [ "issue:23647|issue:29870", - "issue:43381|issue:43701", + "issue:44871|issue:44977", "issue:44361|issue:45198", "issue:43278|issue:43575", "issue:29127|issue:34567", - "issue:41720|issue:45229", "issue:46421|issue:46858", "issue:44206|issue:44479", + "issue:41720|issue:45229", "issue:43412|issue:45588", "issue:44265|issue:45198", "issue:43653|issue:44442", "issue:43611|issue:46121", + "issue:29870|issue:34689", "issue:43976|issue:45440", "issue:45198|issue:45803", - "issue:29870|issue:34689", - "issue:44977|issue:45563", - "issue:43653|issue:44743", "issue:44077|issue:47276", + "issue:44977|issue:45563", "issue:43334|issue:43531", + "issue:43653|issue:44743", "issue:42915|issue:42994", "issue:43065|issue:43994", - "issue:43756|issue:45440", - "issue:44568|issue:45491", "issue:43575|issue:43701", + "issue:43756|issue:45440", "issue:43575|issue:44387", + "issue:44568|issue:45491", "issue:42490|issue:43381", "issue:29127|issue:33357", "issue:43994|issue:45412", "issue:43065|issue:45412", - "issue:45491|issue:46693", "issue:44442|issue:45072", "issue:45372|issue:45588" ], - "prepared_review_unit_hash": "93601e0eeefc7c9d2468e9481424d01ac38a240fd4d7a3b937f65b6500143af7", + "prepared_review_unit_hash": "4ad3e3d572b8d1a7f6b79f7f75523ce7f165bf68a782144fb32140e3443cc157", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37386, - "estimated_input_tokens": 9347, - "estimated_eval_tokens": 18950 + "soft_pair_count": 31, + "serialized_chars": 37124, + "estimated_input_tokens": 9281, + "estimated_eval_tokens": 18818 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the issues span unrelated bugs in tokenizers, generation, model loading, quantization, export, cache handling, and model-specific regressions. None of the suggested soft pairs look like the same underlying defect.", - "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue fits this cluster; the items describe many unrelated problems across different models and subsystems.", - "canonical_pr_reason": "No PRs are present in this cluster.", - "best_issue_reason": "No global best issue: none of these issues subsumes the others or represents the shared bug for the cluster.", - "best_pr_reason": "No PRs are present in this cluster.", + "summary": "This cluster is a grab-bag of unrelated Transformers issues across tokenizers, loading, generation, export, cache behavior, and model-specific regressions. The soft-similarity links are mostly superficial, so none of the candidate pairs should be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue: the items span unrelated bug classes and code paths, so there is no true duplicate cluster anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44442 is the most self-contained and reproducible standalone bug report in the set, so it is the best representative if one issue must be picked, but it does not unify the cluster.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:23647", "right": "issue:29870", "accept": false, - "reason": "Unrelated: Wandb sweeps custom objective vs BART generate() decoder_attention_mask UnboundLocalError." + "reason": "Unrelated bugs: W&B sweeps custom objective integration vs BART generate() decoder_attention_mask initialization error." }, { - "left": "issue:43381", - "right": "issue:43701", + "left": "issue:44871", + "right": "issue:44977", "accept": false, - "reason": "Different problems: eval-mode gradient checkpointing restriction vs resume_from_checkpoint key mismatch." + "reason": "Different model paths and failures: Gemma eos_token_id config mismatch vs Qwen3.5 flash-attention generation bug." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "Both tokenizer-related, but different models and failures: MLukeTokenizer AttributeError vs Wav2Vec2 save/tokenization issues." + "reason": "Different tokenizers/models and symptoms: MLuke task AttributeError vs Wav2Vec2 save_pretrained/tokenization failure." }, { "left": "issue:43278", "right": "issue:43575", "accept": false, - "reason": "Different symptoms and code paths: dtype drift in evaluate vs tensor-parallel loading OOM." + "reason": "Both mention loading/training behavior, but one is dtype drift in eval and the other is tensor-parallel OOM on model load." }, { "left": "issue:29127", "right": "issue:34567", "accept": false, - "reason": "LayoutLMv3 box-info error is unrelated to TrainerState num_input_tokens_seen not updating." - }, - { - "left": "issue:41720", - "right": "issue:45229", - "accept": false, - "reason": "Both are inference failures, but one is CUDA assert on Qwen3 auto device map and the other is Gemma4 multi-GPU OOM." + "reason": "LayoutLMv3 error messaging vs TrainerState token-count tracking are unrelated code paths." }, { "left": "issue:46421", "right": "issue:46858", "accept": false, - "reason": "Cache-related, but distinct bugs: spurious recompilation with chunked prefill vs device-tensor seq_length breaking slice arithmetic." + "reason": "Both involve cache/compile behavior, but one is lazy cache init causing recompilation and the other is a device-tensor seq-length arithmetic bug." }, { "left": "issue:44206", "right": "issue:44479", "accept": false, - "reason": "Different paths: LasrFeatureExtractor unsupported center arg vs video-input regression in Qwen VL models." + "reason": "Different regressions: unsupported feature-extractor arg vs video-input regression across Qwen variants." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Both are OOM/assertion-type failures, but they affect different models and likely different memory/device paths." }, { "left": "issue:43412", "right": "issue:45588", "accept": false, - "reason": "Executorch export failure in Gemma3n is unrelated to flash_attention s_aux=None crash." + "reason": "Export-time Gemma3n issues vs flash-attention runtime crash with s_aux=None are distinct failures." }, { "left": "issue:44265", "right": "issue:45198", "accept": false, - "reason": "torch.export failure for torch_compilable_check is unrelated to Wav2Vec2 save_pretrained/tokenization failure." + "reason": "torch.export.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failure." }, { "left": "issue:43653", "right": "issue:44442", "accept": false, - "reason": "Tokenizer bugs, but not the same: BigBird special-token registration vs AutoTokenizer failing to load FastSpeech2ConformerTokenizer." + "reason": "Tokenizer special-token registration bug vs AutoTokenizer loading failure for a different tokenizer/model." }, { "left": "issue:43611", "right": "issue:46121", "accept": false, - "reason": "Different regressions: base_model_prefix loading break vs convert_rope_params_to_dict TypeError on JSON list input." + "reason": "Model loading with base_model_prefix vs rope-validation helper TypeError are separate mechanisms." + }, + { + "left": "issue:29870", + "right": "issue:34689", + "accept": false, + "reason": "BART generation decoder mask bug vs Llama 3.2 Vision loading break are different model/version regressions." }, { "left": "issue:43976", "right": "issue:45440", "accept": false, - "reason": "Unrelated: Python version compatibility regression vs DeepSeekV3 native/remote implementation divergence." + "reason": "Python version compatibility issue vs DeepSeekV3 implementation divergence are unrelated." }, { "left": "issue:45198", "right": "issue:45803", "accept": false, - "reason": "Different components: Wav2Vec2 save/tokenization failure vs FuyuBatchFeature bare-except swallowing errors." + "reason": "Wav2Vec2 tokenization/save failure vs bare-except swallowing interrupts in FuyuBatchFeature are different bugs." }, { - "left": "issue:29870", - "right": "issue:34689", + "left": "issue:44077", + "right": "issue:47276", "accept": false, - "reason": "BART generation UnboundLocalError is unrelated to Llama 3.2 Vision model loading breakage." + "reason": "PatchTSMixer post_init policy change vs synced_gpus warning wording are unrelated." }, { "left": "issue:44977", "right": "issue:45563", "accept": false, - "reason": "Flash-attention generation regression vs stale warning in paged generate(); different bug and severity." - }, - { - "left": "issue:43653", - "right": "issue:44743", - "accept": false, - "reason": "BigBird tokenizer special-token issue is unrelated to Qwen3.5 recurrent-state reset in cache mode." + "reason": "Generation failure on Qwen3.5 vs stale warning in paged generate() are different issues." }, { - "left": "issue:44077", - "right": "issue:47276", + "left": "issue:43334", + "right": "issue:43531", "accept": false, - "reason": "PatchTST post_init policy vs synced_gpus warning message; no shared code-path bug." + "reason": "Qwen3-VL missing pad_token_id and Qwen3-MoE sliding_window behavior are separate model-specific bugs." }, { - "left": "issue:43334", - "right": "issue:43531", + "left": "issue:43653", + "right": "issue:44743", "accept": false, - "reason": "Qwen3-VL pad_token_id load failure is unrelated to Qwen3-MoE sliding_window handling." + "reason": "BigBird tokenizer special-token registration vs Qwen3 recurrent-state reset are unrelated." }, { "left": "issue:42915", "right": "issue:42994", "accept": false, - "reason": "Both mention quantization, but one is FP8 config failure and the other is quantized model saving failure." + "reason": "Qwen3Moe FP8 config failure vs quantized model saving failure are different code paths." }, { "left": "issue:43065", "right": "issue:43994", "accept": false, - "reason": "Different models and failures: Sam3PixelDecoder dummy conv vs Siglip2 nonsensical outputs." + "reason": "Dummy Conv2d in Sam3PixelDecoder vs Siglip2 nonsensical outputs are unrelated model problems." }, { - "left": "issue:43756", - "right": "issue:45440", + "left": "issue:43575", + "right": "issue:43701", "accept": false, - "reason": "Smollm3 RoPE layer mismatch is unrelated to DeepSeekV3MoE divergence." + "reason": "Tensor-parallel OOM on Qwen2 load vs resume_from_checkpoint key mismatch are different failure modes." }, { - "left": "issue:44568", - "right": "issue:45491", + "left": "issue:43756", + "right": "issue:45440", "accept": false, - "reason": "Tokenizer special-token addition bug is unrelated to Gemma3 NaN embeddings with sliding-window attention." + "reason": "Smollm3 RoPE-layer discrepancy vs DeepSeekV3 divergence are unrelated architecture/config issues." }, { "left": "issue:43575", - "right": "issue:43701", + "right": "issue:44387", "accept": false, - "reason": "Tensor-parallel OOM is unrelated to resume_from_checkpoint key mismatch." + "reason": "Both are memory complaints, but one is TP load OOM and the other is increased reserved memory under int4 quantization." }, { - "left": "issue:43575", - "right": "issue:44387", + "left": "issue:44568", + "right": "issue:45491", "accept": false, - "reason": "Both are OOMs, but one is TP load OOM and the other is increased reserved memory under int4 quantization; not the same bug." + "reason": "Tokenizer add_special_tokens regression vs Gemma3 sliding-window NaN embeddings are different bugs." }, { "left": "issue:42490", "right": "issue:43381", "accept": false, - "reason": "Local-path trailing slash import warning is unrelated to eval-mode gradient checkpointing restriction." + "reason": "Local-path import warning and gradient-checkpointing-in-eval are unrelated." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "LayoutLMv3 box-info error and MacOS CLIP bus error are unrelated model-specific issues." + "reason": "LayoutLMv3 error clarity vs CLIP bus error on macOS are unrelated." }, { "left": "issue:43994", "right": "issue:45412", "accept": false, - "reason": "Siglip2 wrong outputs vs RT-DETR memory not released; different failure modes and code paths." + "reason": "Siglip2 loading/output issue vs RT-DETR memory leak are different model classes and symptoms." }, { "left": "issue:43065", "right": "issue:45412", "accept": false, - "reason": "Sam3PixelDecoder dummy Conv2d and RT-DETR GC/memory leak are unrelated." - }, - { - "left": "issue:45491", - "right": "issue:46693", - "accept": false, - "reason": "Gemma3 NaN embeddings from sliding-window padding is unrelated to FA2 long-context performance regression." + "reason": "Sam3PixelDecoder dummy layer vs RT-DETR deletion/GC memory leak are unrelated." }, { "left": "issue:44442", "right": "issue:45072", "accept": false, - "reason": "Tokenizer loading failure vs dtype mismatch in bf16 inference; different subsystems." + "reason": "AutoTokenizer loading failure vs dtype-mismatch inference bug in SwitchTransformers/TimmWrapperModel are different code paths." }, { "left": "issue:45372", "right": "issue:45588", "accept": false, - "reason": "Gemma4 processor import issue is unrelated to flash_attention crash on sink-less models." + "reason": "Gemma 4 processor import failure from mistral_common vs flash_attention.py crash on sink-less models are unrelated." } ] }, @@ -11292,12 +11235,11 @@ "issue:43381", "issue:43421", "issue:43526", - "issue:43646", "issue:43824", "issue:43873", "issue:43931", "issue:43994", - "issue:44077", + "issue:44186", "issue:44279", "issue:44442", "issue:44462", @@ -11328,251 +11270,245 @@ "issue:47217" ], "soft_pairs": [ - "issue:45406|issue:46701", + "issue:45491|issue:46693", "issue:43526|issue:47217", + "issue:45406|issue:46701", "issue:44625|issue:45584", "issue:29127|issue:35141", "issue:45440|issue:45593", + "issue:45459|issue:45712", "issue:43421|issue:45491", "issue:43526|issue:45834", "issue:45250|issue:46640", - "issue:45459|issue:45712", - "issue:45412|issue:46369", "issue:45372|issue:45406", + "issue:45412|issue:46369", "issue:43824|issue:44279", - "issue:45245|issue:45563", "issue:33290|issue:34689", "issue:33290|issue:35141", "issue:43381|issue:43873", "issue:27957|issue:33290", + "issue:45245|issue:45563", "issue:45372|issue:45459", + "issue:43526|issue:44186", "issue:43994|issue:46369", - "issue:43526|issue:46693", - "issue:42915|issue:43931", + "issue:44857|issue:45588", "issue:23647|issue:28282", + "issue:42915|issue:43931", "issue:43066|issue:44462", - "issue:44857|issue:45588", "issue:29870|issue:29925", "issue:45588|issue:46821", "issue:27957|issue:29127", "issue:44442|issue:44898", - "issue:44514|issue:45381", - "issue:43646|issue:44077" + "issue:44514|issue:45381" ], - "prepared_review_unit_hash": "783caca32cf2e134d990bff29617161dd83c0885cf7a2032efa515ee0fd3a3a3", + "prepared_review_unit_hash": "d2c47051c532ce9590f5cedbddc6bd026b41862578d685b71d70723bdc38dab1", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36460, - "estimated_input_tokens": 9115, - "estimated_eval_tokens": 18486 + "serialized_chars": 35929, + "estimated_input_tokens": 8983, + "estimated_eval_tokens": 18222 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: the soft-similar issue pairs mostly share only broad Transformers vocabulary (e.g. model names, tokenizers, Flash Attention, generation) but describe different bugs, different code paths, or different models. I would not merge any of the proposed soft edges.", - "confidence": 0.95, - "canonical_issue_reason": "No single issue is a safe canonical here; the items span unrelated bugs across many subsystems, so any choice would be arbitrary.", + "summary": "This cluster is a loose mix of unrelated Transformers issues spanning tokenizers, vision processors, generation, Flash Attention, quantization, and model-loading. None of the soft pairs look like the same underlying bug, so they should stay separate.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical duplicate is evident; if one issue must anchor the cluster, issue 43873 is the broadest/highest-activity open item, but it does not actually subsume the others.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong global representative of this cluster because the similarity is mostly topical rather than duplicate-level.", + "best_issue_reason": "Issue 43873 is the best cluster anchor because it is active, open, and broad ('offloading' + quantization), but it is still not a good representative duplicate for the mixed set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45406", - "right": "issue:46701", + "left": "issue:45491", + "right": "issue:46693", "accept": false, - "reason": "Both touch Gemma-related code, but one is a missing processor attribute and the other is an image-text pipeline failure; different bugs." + "reason": "Different failures: NaN embeddings in mixed-length batching vs FA2 long-context performance regression." }, { "left": "issue:43526", "right": "issue:47217", "accept": false, - "reason": "Unrelated areas: BeitImageProcessorFast label reduction vs VL image-placeholder handling." + "reason": "Different subsystems: BeitImageProcessorFast reduce_labels vs VL image-placeholder handling in user text." + }, + { + "left": "issue:45406", + "right": "issue:46701", + "accept": false, + "reason": "Different Gemma-related bugs: missing _tokenizer in serve vs image-text pipeline output issue." }, { "left": "issue:44625", "right": "issue:45584", "accept": false, - "reason": "Different models and failures: Qwen3.5 config propagation vs Whisper empty-transcription generation." + "reason": "Different models and code paths: Qwen3.5 num_labels propagation vs Whisper empty-transcription failure after align_special_tokens." }, { "left": "issue:29127", "right": "issue:35141", "accept": false, - "reason": "LayoutLMv3 error messaging is unrelated to embedding reinitialization after resize." + "reason": "Different concerns: LayoutLMv3 error-message clarity vs embedding resize reinitialization bug." }, { "left": "issue:45440", "right": "issue:45593", "accept": false, - "reason": "Different model implementations and symptoms: DeepseekV3MoE divergence vs D-FINE auxiliary-loss behavior." + "reason": "Different models/features: DeepseekV3MoE remote divergence vs D-FINE auxiliary-loss omission." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Different issues: protobuf exception masking tokenizer errors vs leftover dummy classes in dummy_pt_objects.py." }, { "left": "issue:43421", "right": "issue:45491", "accept": false, - "reason": "TokenizersBackend post-processor updates and Gemma3 NaN embeddings are separate code paths." + "reason": "Different bugs: tokenizer post-processor not updating runtime special tokens vs Gemma3 NaN embeddings." }, { "left": "issue:43526", "right": "issue:45834", "accept": false, - "reason": "Different processors and failures: Beit label handling vs Kosmos2.5 OCR index error." + "reason": "Different issues: Beit reduce_labels returning one label vs Kosmos2.5 long OCR input index error." }, { "left": "issue:45250", "right": "issue:46640", "accept": false, - "reason": "Both are generic Flash Attention titles, but there is no evidence they describe the same concrete bug." + "reason": "Too vague, but these are still separate FA2 reports with different scope and symptoms." }, { - "left": "issue:45459", - "right": "issue:45712", + "left": "issue:45372", + "right": "issue:45406", "accept": false, - "reason": "Tokenizer protobuf error handling is unrelated to dummy class cleanup and dir(transformers) leakage." + "reason": "Different bugs: ReasoningEffort import error breaking Gemma4 loading vs Gemma4Processor _tokenizer attribute error in serve." }, { "left": "issue:45412", "right": "issue:46369", "accept": false, - "reason": "Both mention memory, but RT-DETR GC cleanup and SAM3 LiteText overflow are different failures." - }, - { - "left": "issue:45372", - "right": "issue:45406", - "accept": false, - "reason": "Gemma 4 processor loading fails for different reasons: mistral_common import vs missing _tokenizer." + "reason": "Different memory problems: RT-DETR GC leak vs SAM3-litetext memory overflow." }, { "left": "issue:43824", "right": "issue:44279", "accept": false, - "reason": "The first is a specific missing export import error; the second is only a broad dependency issue." - }, - { - "left": "issue:45245", - "right": "issue:45563", - "accept": false, - "reason": "Category-count runtime error and a stale warning in paged generate are unrelated." + "reason": "Different issues: Qwen2.5-VL import failure vs generic dependency issue with transformers." }, { "left": "issue:33290", "right": "issue:34689", "accept": false, - "reason": "Adafactor/deepspeed OOM is unrelated to Llama 3.2 Vision model loading breakage." + "reason": "Different bugs: deepspeed Adafactor OOM vs Llama 3.2 Vision model-loading regression." }, { "left": "issue:33290", "right": "issue:35141", "accept": false, - "reason": "Different subsystems: optimizer memory use vs embedding resize/post_init behavior." + "reason": "Different bugs: optimizer OOM vs embedding resize reinitialization." }, { "left": "issue:43381", "right": "issue:43873", "accept": false, - "reason": "Gradient checkpointing in eval mode and quantized offloading are separate issues." + "reason": "Different behaviors: eval-mode gradient-checkpointing prohibition vs quantization offloading failure." }, { "left": "issue:27957", "right": "issue:33290", "accept": false, - "reason": "XLMRoberta Flash Attention 2 and deepspeed OOM are not the same bug." + "reason": "Different models and failures: XLMRoberta Flash Attention 2 vs deepspeed Adafactor OOM." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Different issues: category-count limit vs stale generate warning for num_return_sequences." }, { "left": "issue:45372", "right": "issue:45459", "accept": false, - "reason": "Gemma processor import dependency failure and protobuf tokenizer error masking are unrelated." + "reason": "Different bugs: import error for ReasoningEffort vs protobuf-specific tokenizer error masking." }, { - "left": "issue:43994", - "right": "issue:46369", + "left": "issue:43526", + "right": "issue:44186", "accept": false, - "reason": "SigLIP2 incorrect outputs and SAM3 LiteText memory overflow are different problems." + "reason": "Different tokenizer/image-processor bug vs LayoutLMv2Tokenizer NER padding/truncation crash." }, { - "left": "issue:43526", - "right": "issue:46693", + "left": "issue:43994", + "right": "issue:46369", "accept": false, - "reason": "Beit label reduction and FA2 long-context regression are unrelated." + "reason": "Different models and symptoms: SigLIP2 wrong outputs vs SAM3 memory overflow." }, { - "left": "issue:42915", - "right": "issue:43931", + "left": "issue:44857", + "right": "issue:45588", "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen3-VL weight-shape mismatch are different load/config bugs." + "reason": "Different code paths: RT-DETR image loss AMP crash vs flash_attention sink-less AttributeError." }, { "left": "issue:23647", "right": "issue:28282", "accept": false, - "reason": "Wandb sweeps custom objective and missing PyTorch import are unrelated." + "reason": "Different areas: wandb sweeps custom objective vs missing PyTorch ImportError." }, { - "left": "issue:43066", - "right": "issue:44462", + "left": "issue:42915", + "right": "issue:43931", "accept": false, - "reason": "Wrong decoder type in v5 and ignoring tokenizer.json are different tokenizer-loading issues." + "reason": "Different Qwen model failures: FineGrainedFP8Config failure vs Qwen3-VL weight-shape mismatch." }, { - "left": "issue:44857", - "right": "issue:45588", + "left": "issue:43066", + "right": "issue:44462", "accept": false, - "reason": "LwDetrImageLoss AMP crash and flash_attention.py s_aux=None crash are different code paths." + "reason": "Related tokenizer area, but different concrete bugs: wrong decoder type in v5 vs ignoring tokenizer.json." }, { "left": "issue:29870", "right": "issue:29925", "accept": false, - "reason": "BART decoder_attention_mask bug and SigLIP fast tokenizer enforcement are unrelated." + "reason": "Different issues: BART generate decoder-mask bug vs SigLIP fast-tokenizer enforcement." }, { "left": "issue:45588", "right": "issue:46821", "accept": false, - "reason": "Flash attention sinkless-model crash and TimesFM window_size AttributeError are different failures." + "reason": "Different subsystems: flash_attention sink-less crash vs TimesFM window_size AttributeError." }, { "left": "issue:27957", "right": "issue:29127", "accept": false, - "reason": "No shared bug: XLMRoberta FA2 support vs LayoutLMv3 error clarity." + "reason": "Different models and code paths: XLMRoberta Flash Attention 2 vs LayoutLMv3 error-message clarity." }, { "left": "issue:44442", "right": "issue:44898", "accept": false, - "reason": "AutoTokenizer loading failure and Perceiver interpolate_pos_encoding failure are different models and paths." + "reason": "Different failures: tokenizer loading vs Perceiver non-default-resolution failure." }, { "left": "issue:44514", "right": "issue:45381", "accept": false, - "reason": "Both are Qwen2.5-VL, but one is batched chat-template padding and the other is video vision_position_ids." - }, - { - "left": "issue:43646", - "right": "issue:44077", - "accept": false, - "reason": "Generic custom model init breakage is broader than the specific patchtsmixer optional post_init change." - }, - { - "left": "issue:43873", - "right": "issue:45412", - "accept": false, - "reason": "Quantization offloading and RT-DETR object lifetime/memory release are unrelated." + "reason": "Same model family, but different concrete bugs: batched apply_chat_template crash vs wrong video vision_position_ids." } ] }, @@ -11611,6 +11547,7 @@ "issue:43575", "issue:43606", "issue:43643", + "issue:43646", "issue:43701", "issue:43720", "issue:43742", @@ -11619,18 +11556,19 @@ "issue:43827", "issue:43873", "issue:43931", - "issue:43950", "issue:44077", "issue:44206", "issue:44442", "issue:44479", "issue:44743", "issue:44811", + "issue:44855", "issue:44871", "issue:44913", "issue:44964", "issue:44977", "issue:44987", + "issue:45137", "issue:45245", "issue:45440", "issue:45759", @@ -11641,221 +11579,227 @@ "issue:46399", "issue:46821", "issue:47312", - "issue:47427", "issue:47436" ], "soft_pairs": [ + "issue:43646|issue:44077", + "issue:46821|issue:47436", + "issue:43316|issue:47312", "issue:46121|issue:46399", + "issue:43262|issue:44206", "issue:43761|issue:45859", - "issue:43316|issue:47312", - "issue:46821|issue:47436", "issue:43720|issue:44442", - "issue:43262|issue:44206", "issue:25251|issue:28282", "issue:44743|issue:44811", - "issue:44977|issue:45245", "issue:42175|issue:46234", "issue:44077|issue:44479", + "issue:44977|issue:45245", "issue:43873|issue:44871", "issue:42490|issue:44743", "issue:43316|issue:45759", "issue:43493|issue:43827", - "issue:43749|issue:43931", "issue:42617|issue:43575", + "issue:43749|issue:43931", "issue:43643|issue:44913", - "issue:43550|issue:43701", "issue:44987|issue:46178", + "issue:43550|issue:43701", "issue:27957|issue:29870", "issue:43316|issue:43525", "issue:43606|issue:43701", "issue:44964|issue:45440", "issue:43575|issue:43742", - "issue:43950|issue:47427" + "issue:44855|issue:45137" ], - "prepared_review_unit_hash": "cf79c048e247257c59fa4ce714a30327c76171931a153bd3322273a7602c0a6e", + "prepared_review_unit_hash": "dcb5c63d8c10f3e3720cb68611e8325534d7109a2555acb9f873d08c4e1519df", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 26, - "serialized_chars": 34948, - "estimated_input_tokens": 8737, - "estimated_eval_tokens": 17730 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35789, + "estimated_input_tokens": 8948, + "estimated_eval_tokens": 18152 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are largely unrelated one-off transformer bugs across different models, configs, and subsystems. None of the soft pairs appear to describe the same concrete underlying issue closely enough to merge as duplicates.", - "confidence": 0.22, - "canonical_issue_reason": "Issue 43873 is the broadest and most active loading/offload-related report in the set, with the strongest discussion/references footprint; it is the best single anchor only in a loose sense.", + "summary": "This set is mostly heterogeneous regression/bug reports with only superficial similarity by subsystem or wording. None of the soft pairs look like the same underlying bug/change, so no duplicate merges are warranted.", + "confidence": 0.94, + "canonical_issue_reason": "No clear canonical issue: the soft pairs are mostly unrelated, and the closest match (issue:43646) is only a broad regression report rather than a true duplicate hub.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43873 is the most central and reusable representative of the cluster because it targets a broad runtime/loading path rather than a model-specific edge case.", + "best_issue_reason": "Issue:43646 is the closest thing to an umbrella report because it is broad and regression-like, but the cluster is too mixed to collapse into one duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:46121", - "right": "issue:46399", + "left": "issue:43646", + "right": "issue:44077", "accept": false, - "reason": "Different failures: rope-parameter validation TypeError vs Qwen3.5 kernel registration/decorator crash." + "reason": "Both touch custom model init, but one is a broad v5 regression while the other is a patchtsmixer-specific post_init rule; not the same bug." }, { - "left": "issue:43761", - "right": "issue:45859", + "left": "issue:46821", + "right": "issue:47436", "accept": false, - "reason": "Different models and code paths: CLIPVision hidden_states regression vs Qwen3.5 MoE rotary embedding CPU-offload incompatibility." + "reason": "Different models and different missing attributes (`window_size` vs `max_position_embeddings`); same error class, not same issue." }, { "left": "issue:43316", "right": "issue:47312", "accept": false, - "reason": "One is a Gemma3 config API mismatch; the other is a generic kernels 0.16 compatibility issue." + "reason": "Gemma3TextConfig API mismatch and kernels 0.16 incompatibility are unrelated problems." }, { - "left": "issue:46821", - "right": "issue:47436", + "left": "issue:46121", + "right": "issue:46399", "accept": false, - "reason": "Unrelated config attributes: TimesFM window_size vs Qwen2.5 Omni max_position_embeddings." + "reason": "Rope-parameter JSON/list handling is a different code path from the Qwen3.5 kernel registration crash." }, { - "left": "issue:43720", - "right": "issue:44442", + "left": "issue:43262", + "right": "issue:44206", "accept": false, - "reason": "Packed-weight accelerate loading bug for BitNet vs tokenizer loading failure for FastSpeech2Conformer." + "reason": "Audio chat-template sampling-rate default and unsupported `center` arg in a feature extractor are distinct audio bugs." }, { - "left": "issue:43262", - "right": "issue:44206", + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "CLIPVision hidden-state regression and Qwen3.5 MoE CPU-offload incompatibility are different failures." + }, + { + "left": "issue:43720", + "right": "issue:44442", "accept": false, - "reason": "Audio processor chat-template sampling-rate default vs feature extractor unsupported center argument." + "reason": "BitNet packed-weight loading and FastSpeech2Conformer tokenizer loading are unrelated model-loading problems." }, { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Pipeline output nesting regression is unrelated to a missing PyTorch import." + "reason": "Pipeline `top_k` output shape bug and missing PyTorch ImportError are completely different." }, { "left": "issue:44743", "right": "issue:44811", "accept": false, - "reason": "Different behaviors in different modules: recurrent cache reset vs Whisper batch_decode special-token handling." - }, - { - "left": "issue:44977", - "right": "issue:45245", - "accept": false, - "reason": "Qwen3.5 flash-attention generation bug is unrelated to the category-count limit error." + "reason": "Qwen3.5 recurrent-state reset and Whisper `batch_decode` special-token handling are separate behaviors." }, { "left": "issue:42175", "right": "issue:46234", "accept": false, - "reason": "Packaging/backend install issue vs missing import_utils helper; not the same bug." + "reason": "Packaging/backend omission and a missing utility symbol are not the same bug." }, { "left": "issue:44077", "right": "issue:44479", "accept": false, - "reason": "PatchTST mixer post_init API constraint is unrelated to Qwen video-input regression." + "reason": "PatchTSMixer `post_init` validation and Qwen video-input regression do not share a concrete code path." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Flash-attention generation issues and category-count overflow are unrelated." }, { "left": "issue:43873", "right": "issue:44871", "accept": false, - "reason": "Quantization/offloading behavior bug vs Gemma eos_token_id config inconsistency." + "reason": "Quantized offloading behavior and Gemma eos-token config inconsistency are different issues." }, { "left": "issue:42490", "right": "issue:44743", "accept": false, - "reason": "Local path parsing/import warning is unrelated to Qwen recurrent-state reset." + "reason": "Local-path import warnings and Qwen recurrent-state handling are unrelated." }, { "left": "issue:43316", "right": "issue:45759", "accept": false, - "reason": "Both are config-related, but they concern different models and distinct API mismatches." + "reason": "Both are config-related, but one is a Gemma3TextConfig API discrepancy and the other is a multimodal `text_config` unwrapping bug." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation divergence is unrelated to docs still mentioning pipeline()." + "reason": "SigLIP2 implementation drift and outdated docs referencing `pipeline()` are not the same underlying change." }, { - "left": "issue:43749", - "right": "issue:43931", + "left": "issue:42617", + "right": "issue:43575", "accept": false, - "reason": "Different loading failures: FSDP CPU RAM efficient loading vs Qwen3-VL weight-shape mismatch." + "reason": "3d_parallel runtime failure and tensor-parallel OOM are different loading/execution problems." }, { - "left": "issue:42617", - "right": "issue:43575", + "left": "issue:43749", + "right": "issue:43931", "accept": false, - "reason": "3d_parallel.py execution failure is not the same as Qwen2-57B tensor-parallel OOM." + "reason": "FSDP CPU-RAM-efficient loading breakage and Qwen3-VL shape mismatch are separate model-loading bugs." }, { "left": "issue:43643", "right": "issue:44913", "accept": false, - "reason": "Missing fields from trust_remote_code config loading vs GPTNeoX rotary_pct reset on reload." + "reason": "Remote-code config field loss and GPTNeoX rotary_pct not persisting on reload are different config persistence issues." }, { - "left": "issue:43550", - "right": "issue:43701", + "left": "issue:44987", + "right": "issue:46178", "accept": false, - "reason": "torch.compile/SDPA failure is unrelated to resume_from_checkpoint key mismatch." + "reason": "Transformers>=5.1 loading failure and tensorflow_text backend registration are unrelated." }, { - "left": "issue:44987", - "right": "issue:46178", + "left": "issue:43550", + "right": "issue:43701", "accept": false, - "reason": "Model load failure for physical-intelligence/fast is unrelated to tensorflow_text backend registration." + "reason": "torch.compile/SDPA failure in Bamba and resume checkpoint key mismatch do not share a code path." }, { "left": "issue:27957", "right": "issue:29870", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta is unrelated to Bart decoder_attention_mask assignment." + "reason": "Flash Attention 2 support for XLMRoberta and BART generate UnboundLocalError are unrelated." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Different config attributes and models: Gemma3TextConfig API discrepancy vs Llama4Config pad_token_id missing." + "reason": "Both involve config APIs, but the missing/variant fields are model-specific and not the same bug." }, { "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "CPU offload device mismatch is unrelated to resume_from_checkpoint key mismatch." + "reason": "CPU offload device mismatch and checkpoint-key mismatch are different loading errors." }, { "left": "issue:44964", "right": "issue:45440", "accept": false, - "reason": "Phi-4 multimodal load regression and DeepseekV3MoE divergence are different model-specific issues." + "reason": "Phi-4 multimodal loading failure and DeepSeekV3MoE divergence from remote implementation are not the same problem." }, { "left": "issue:43575", "right": "issue:43742", "accept": false, - "reason": "Tensor-parallel OOM on a large Qwen model is not the same as a key error loading MobileLLM-125M." + "reason": "Tensor-parallel OOM and MobileLLM key error are distinct model-loading issues." }, { - "left": "issue:43950", - "right": "issue:47427", + "left": "issue:44855", + "right": "issue:45137", "accept": false, - "reason": "Non-persistent buffer corruption on load is a different regression from redundant init skipping on non-CUDA hardware." + "reason": "Python 3.13 import parsing failure and DeepSpeed ZeRO3 deque underflow are unrelated runtime bugs." } ] }, @@ -11890,7 +11834,6 @@ "issue:43232", "issue:43295", "issue:43316", - "issue:43352", "issue:43388", "issue:43452", "issue:43646", @@ -11899,21 +11842,21 @@ "issue:43761", "issue:43825", "issue:43931", + "issue:43950", "issue:44062", "issue:44077", "issue:44079", "issue:44188", "issue:44190", "issue:44442", + "issue:44464", "issue:44496", "issue:44521", "issue:44534", "issue:44589", "issue:44617", - "issue:44855", "issue:44871", "issue:44913", - "issue:45137", "issue:45216", "issue:45584", "issue:45803", @@ -11921,232 +11864,233 @@ "issue:46050", "issue:46234", "issue:46291", + "issue:46421", "issue:46506", - "issue:46640", "issue:46829", "issue:46932", "issue:47217", - "issue:47276" + "issue:47276", + "issue:47427" ], "soft_pairs": [ + "issue:43950|issue:47427", "issue:44062|issue:45907", "issue:43761|issue:44079", - "issue:44855|issue:45137", "issue:43701|issue:46506", "issue:44190|issue:44496", "issue:43316|issue:43716", "issue:43931|issue:45216", - "issue:44442|issue:45803", "issue:46829|issue:46932", "issue:27957|issue:32090", + "issue:44442|issue:45803", "issue:44589|issue:47276", "issue:43295|issue:44871", "issue:23647|issue:28218", "issue:44871|issue:45584", "issue:43295|issue:43825", - "issue:43761|issue:44188", "issue:16998|issue:22355", - "issue:43232|issue:43388", + "issue:43761|issue:44188", "issue:46234|issue:46291", - "issue:44521|issue:47217", + "issue:43232|issue:43388", "issue:43452|issue:44496", + "issue:44521|issue:47217", "issue:44077|issue:44534", "issue:43646|issue:44913", "issue:44871|issue:46050", "issue:25251|issue:27957", "issue:43065|issue:44617", - "issue:43352|issue:46640" + "issue:44464|issue:46421" ], - "prepared_review_unit_hash": "73643fdd8e7fc745b9d341fdba4a3d2cecc38fba5d0f562ee1e4cbc2e0ef0d6e", + "prepared_review_unit_hash": "77e41233528be6393571f89e132bd232764a6ae408e2a7cb472a3e1d8e785a5f", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35665, - "estimated_input_tokens": 8917, - "estimated_eval_tokens": 18090 + "serialized_chars": 35741, + "estimated_input_tokens": 8936, + "estimated_eval_tokens": 18128 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The candidate pairs are largely unrelated despite some surface similarity in model/tokenizer/generation wording. I would not merge any of these soft edges; they describe distinct bugs or feature requests in different code paths. No single canonical issue stands out for the cluster.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "The cluster is mostly a grab-bag of unrelated issue reports connected only by broad surface similarity (generation, tokenizers, multimodal, config/loading, etc.). None of the soft pairs look like the same underlying bug/change, so they should not be merged as duplicates.", + "confidence": 0.92, + "canonical_issue_reason": "If a single anchor is needed, #44521 is the strongest representative because it is open, has the most inbound references, and describes a concrete reproducible multimodal regression; however, the overall cluster is too heterogeneous to treat as one duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "#44521 is the best standalone issue in this set: it is active, well-scoped, and has the clearest user-facing reproduction. It does not, however, unify the rest of the issues into one bug.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43950", + "right": "issue:47427", + "accept": false, + "reason": "Both touch initialization behavior, but one is about corrupting non-persistent buffers during load while the other is about skipping redundant parameter init on non-CUDA hardware; different bugs and code paths." + }, { "left": "issue:44062", "right": "issue:45907", "accept": false, - "reason": "Different problems: AddedToken keyword collision vs list_repo_files not forwarding revision. No shared code path." + "reason": "`AddedToken()` argument handling and `list_repo_files` revision forwarding are unrelated API issues in different subsystems." }, { "left": "issue:43761", "right": "issue:44079", "accept": false, - "reason": "Both involve outputs/None handling, but one is CLIPVision hidden_states regression and the other is ModelOutput key assignment. Different bugs." - }, - { - "left": "issue:44855", - "right": "issue:45137", - "accept": false, - "reason": "Unrelated failures: DebertaV2 import parse error on Python 3.13 vs DeepSpeed ZeRO3 deque underflow." + "reason": "One is a CLIPVisionModel regression returning `hidden_states=None`; the other is a `ModelOutput` key assignment bug. Similar surface area, different failure." }, { "left": "issue:43701", "right": "issue:46506", "accept": false, - "reason": "Checkpoint key mismatch is unrelated to CPU/CUDA output differences in PPDocLayoutV3." + "reason": "Checkpoint key mismatches in resume logic and CPU/CUDA output differences in PPDocLayoutV3 are unrelated." }, { "left": "issue:44190", "right": "issue:44496", "accept": false, - "reason": "Local dataset loading in an example script vs model config recognition/loading failure. Different layers and symptoms." + "reason": "Loading a local dataset in a training script and failing to recognize a model config are different loading problems with different causes." }, { "left": "issue:43316", "right": "issue:43716", "accept": false, - "reason": "Gemma3TextConfig API inconsistency is not the same as Mistral-3 image preprocessor dtype mismatch." + "reason": "Gemma3TextConfig API inconsistency and Mistral-3 image preprocessor dtype mismatch are different multimodal issues." }, { "left": "issue:43931", "right": "issue:45216", "accept": false, - "reason": "Model weight shape mismatch during load vs incorrect checkpoint saving behavior. Different directions of the load/save pipeline." - }, - { - "left": "issue:44442", - "right": "issue:45803", - "accept": false, - "reason": "Tokenizer loading failure is unrelated to a bare-except swallowing errors in FuyuBatchFeature.convert_to_tensors()." + "reason": "Both involve Qwen/VL model loading/saving, but one is a weight-shape mismatch on load and the other is an incorrect saved checkpoint in `save_pretrained`; not the same concrete bug." }, { "left": "issue:46829", "right": "issue:46932", "accept": false, - "reason": "VideoPrism hidden_states bug is unrelated to Gemma4 quantized embedder dtype casting." + "reason": "VideoPrism hidden-state field corruption and Gemma4 quantized embedder dtype casting are unrelated model-output vs quantization bugs." }, { "left": "issue:27957", "right": "issue:32090", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta is unrelated to Trainer _gpu_broadcast_one NoneType failure." + "reason": "Flash Attention 2 support for XLMRoberta and a Trainer GPU broadcast `NoneType` error are unrelated." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "FastSpeech2Conformer tokenizer loading and Fuyu batch tensor conversion swallowing exceptions are different tokenizer/data-path issues." }, { "left": "issue:44589", "right": "issue:47276", "accept": false, - "reason": "Float8 storage loading error is unrelated to a misleading synced_gpus warning message." + "reason": "Missing Float8 storage handling and a misleading synced_gpus warning are not the same bug; one is a hard runtime failure, the other is a message issue." }, { "left": "issue:43295", "right": "issue:44871", "accept": false, - "reason": "Custom processor/tokenizer regression is distinct from Gemma-3 eos_token_id inconsistency." + "reason": "A regression in `processor.tokenizer` / image passthrough is unrelated to inconsistent `eos_token_id` configuration." }, { "left": "issue:23647", "right": "issue:28218", "accept": false, - "reason": "Wandb sweeps custom objective and tokenizer added-token spacing are unrelated." + "reason": "WandB sweep custom objective support and tokenizer-added spacing after added tokens have no shared code path." }, { "left": "issue:44871", "right": "issue:45584", "accept": false, - "reason": "Different models and code paths: Gemma-3 eos_token_id config vs Whisper empty transcription after special-token alignment." + "reason": "Both touch generation/tokenizer behavior, but one is an EOS-token config mismatch and the other is empty-transcription failure after special-token alignment; distinct defects." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Processor/tokenizer regression is unrelated to pipeline translation-task error messaging." - }, - { - "left": "issue:43761", - "right": "issue:44188", - "accept": false, - "reason": "CLIPVision hidden_states regression is unrelated to attention-kernel divergence under torch.compile." + "reason": "A custom model API regression and an error-message wording bug in `pipeline()` are not duplicates." }, { "left": "issue:16998", "right": "issue:22355", "accept": false, - "reason": "Model max length question for DeBERTa-V3 is unrelated to missing transformers.onnx import." + "reason": "`model_max_length` behavior in DeBERTa-V3 and a missing `transformers.onnx` module are unrelated." }, { - "left": "issue:43232", - "right": "issue:43388", + "left": "issue:43761", + "right": "issue:44188", "accept": false, - "reason": "Generation kwargs sync issue is unrelated to gather_for_metrics dropping labels in the last batch." + "reason": "Hidden states being dropped in CLIPVisionModel is unrelated to attention-kernel divergence under `torch.compile`." }, { "left": "issue:46234", "right": "issue:46291", "accept": false, - "reason": "Missing import_utils symbol is unrelated to kernels v0.15 requiring version/revision arguments." + "reason": "A missing utility flag in `import_utils` and a kernels import failure due to version/revision requirements are different import-time issues." }, { - "left": "issue:44521", - "right": "issue:47217", + "left": "issue:43232", + "right": "issue:43388", "accept": false, - "reason": "Both involve multimodal behavior, but one is assistant_masks generation and the other is image-placeholder handling in user text. Not the same concrete bug." + "reason": "Generation kwargs updates after `sync_gpus` and `gather_for_metrics` truncating tuple labels are different distributed-training bugs." }, { "left": "issue:43452", "right": "issue:44496", "accept": false, - "reason": "gguf_file loading breakage is unrelated to unrecognized model_type / config.json loading failure." + "reason": "`gguf_file` breaking `from_pretrained` for tokenizer/model loading is unrelated to an unrecognized model config in `AutoModel`." + }, + { + "left": "issue:44521", + "right": "issue:47217", + "accept": false, + "reason": "Both involve multimodal chat/template handling, but one is about assistant masks being all zero and the other is about image placeholders in user text causing model failure; not clearly the same code-path bug." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "Optional post_init policy for patchtsmixer is unrelated to non-persistent buffer initialization junk." + "reason": "Optional `post_init` deprecation/validation and non-persistent buffer corruption are separate initialization problems." }, { "left": "issue:43646", "right": "issue:44913", "accept": false, - "reason": "Generic Transformers 5.0 custom initialization breakage is too broad and not the same as GPTNeoX rotary_pct reload regression." + "reason": "A broad Transformers 5.0 custom model-init breakage and GPTNeoXConfig `rotary_pct` not persisting on reload are different configuration/init bugs." }, { "left": "issue:44871", "right": "issue:46050", "accept": false, - "reason": "Gemma-3 eos_token_id mismatch is unrelated to sinq quantization save/load corruption." + "reason": "`eos_token_id` inconsistency in Gemma-3 and sinq quantized model save/load breakage are different save/load regressions." }, { "left": "issue:25251", "right": "issue:27957", "accept": false, - "reason": "Pipeline output nesting change is unrelated to Flash Attention 2 support." + "reason": "Pipeline top-k nesting behavior and XLMRoberta Flash Attention 2 support are unrelated." }, { "left": "issue:43065", "right": "issue:44617", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as Sam3Video CUDA OOM." + "reason": "A dummy `nn.Conv2d` in Sam3PixelDecoder and Sam3Video CUDA OOM are different SAM3 issues with different failure modes." }, { - "left": "issue:43352", - "right": "issue:46640", + "left": "issue:44464", + "right": "issue:46421", "accept": false, - "reason": "A specific model lacking Flash Attention 2 support is not enough to merge with a generic Flash Attention 2 issue." + "reason": "Chunked generation inconsistencies under compiled forward and lazy cache init causing recompilation are related by theme but not the same concrete bug or fixable in one PR." } ] }, @@ -12177,6 +12121,7 @@ "issue:41950", "issue:42490", "issue:42907", + "issue:43352", "issue:43404", "issue:43441", "issue:43475", @@ -12187,7 +12132,6 @@ "issue:43931", "issue:44060", "issue:44162", - "issue:44464", "issue:44479", "issue:44514", "issue:44534", @@ -12199,252 +12143,251 @@ "issue:45127", "issue:45200", "issue:45381", + "issue:45405", "issue:45542", "issue:45588", "issue:45600", "issue:45632", "issue:45698", "issue:45803", - "issue:46082", "issue:46170", "issue:46178", "issue:46234", "issue:46291", "issue:46365", "issue:46369", - "issue:46421", "issue:46424", + "issue:46640", "issue:46693", - "issue:46697", "issue:47013", "issue:47217" ], "soft_pairs": [ - "issue:44464|issue:46421", "issue:43404|issue:43441", "issue:42907|issue:45698", + "issue:43352|issue:46640", "issue:44162|issue:44792", - "issue:43540|issue:44514", "issue:46178|issue:46291", + "issue:43540|issue:44514", "issue:44792|issue:44977", - "issue:44610|issue:46369", "issue:44479|issue:47217", "issue:45803|issue:46365", - "issue:44534|issue:44913", + "issue:44610|issue:46369", "issue:45381|issue:47217", + "issue:44534|issue:44913", "issue:44792|issue:45600", "issue:45542|issue:46234", "issue:27957|issue:29925", "issue:43525|issue:43531", "issue:43504|issue:45698", - "issue:23647|issue:27957", "issue:35141|issue:42490", + "issue:23647|issue:27957", "issue:46693|issue:47217", "issue:45803|issue:46170", "issue:43931|issue:44060", "issue:45632|issue:46424", - "issue:45588|issue:47013", "issue:44855|issue:45588", - "issue:46082|issue:46697", + "issue:45588|issue:47013", "issue:45127|issue:45200", + "issue:44792|issue:45405", "issue:41950|issue:43475" ], - "prepared_review_unit_hash": "c6159713cbbb1e9e22a0a1ccc6d213049d01c770f940fe64e48458479415df55", + "prepared_review_unit_hash": "20805e9bcfd82a0bcec92e4b0fa7b97f2ff6853b5a2f4a218e7cbdee21c3fbe2", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 36244, - "estimated_input_tokens": 9061, - "estimated_eval_tokens": 18378 + "serialized_chars": 35551, + "estimated_input_tokens": 8888, + "estimated_eval_tokens": 18032 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This batch is heterogeneous: most items are unrelated bugs across different models, processors, and import/cache paths. I found no convincing duplicate pairs among the soft candidates; one pair is superficially close but still not the same concrete bug.", - "confidence": 0.78, - "canonical_issue_reason": "There is no single canonical issue here because the items cover unrelated failures in different subsystems and model families.", + "summary": "This cluster is highly heterogeneous: most items are distinct model- or subsystem-specific bugs, and the soft pairs are mostly superficial text similarity rather than the same underlying defect. I would not merge any of the proposed pairs as duplicates.", + "confidence": 0.88, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 46693 is the strongest standalone report in this set: it describes a clear regression, has broad impact, and received substantial discussion.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44464", - "right": "issue:46421", - "accept": false, - "reason": "Both involve chunked generation/cache behavior, but one is about incorrect outputs and the other about spurious recompilation; different underlying bugs." - }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "Different models and different failure modes: untied lm_head weights vs FlashAttention-related runtime failure." + "reason": "Same model family, but one is tied-weight handling and the other is FlashAttention support; different code paths and fixes." }, { "left": "issue:42907", "right": "issue:45698", "accept": false, - "reason": "One is about saving dequantized Ministral/Devstral models; the other is about loading the wrong custom module after save_pretrained. Related area, not the same bug." + "reason": "Both involve save/load flows, but one is about saving dequantized models and the other about loading the wrong custom module; not the same bug." }, { - "left": "issue:44162", - "right": "issue:44792", + "left": "issue:43352", + "right": "issue:46640", "accept": false, - "reason": "Unrelated model families and symptoms; no shared concrete code-path." + "reason": "Issue 43352 is a specific Nemotron FlashAttention2 support gap; 46640 is a broader FA2 issue with no evidence it is the same failure." }, { - "left": "issue:43540", - "right": "issue:44514", + "left": "issue:44162", + "right": "issue:44792", "accept": false, - "reason": "Both are multimodal input issues, but they affect different processors and different crash conditions." + "reason": "ESM2 breakage and a Janus image-generation test failure are unrelated bugs." }, { "left": "issue:46178", "right": "issue:46291", "accept": false, - "reason": "Distinct dependency/import problems involving different backends and packages." + "reason": "Both are import/dependency failures, but they affect different packages and failure modes." }, { - "left": "issue:44792", - "right": "issue:44977", + "left": "issue:43540", + "right": "issue:44514", "accept": false, - "reason": "Different models and unrelated generation/test failures." + "reason": "Both are multimodal input issues, but they hit different processors/models and different shapes of failure." }, { - "left": "issue:44610", - "right": "issue:46369", + "left": "issue:44792", + "right": "issue:44977", "accept": false, - "reason": "Processor image-size mismatch vs memory overflow; these do not describe the same defect." + "reason": "Janus image generation and Qwen3.5 flash-attention generation problems are unrelated." }, { "left": "issue:44479", "right": "issue:47217", "accept": false, - "reason": "Both are VL-related, but one is a video-input regression and the other is an image-placeholder placement crash; different bugs." + "reason": "Both concern VL inputs, but one is a video-input regression and the other is image-placeholder handling in user text." }, { "left": "issue:45803", "right": "issue:46365", "accept": false, - "reason": "Bare except swallowing errors is unrelated to Trainer import failure from datasets." + "reason": "Bare except swallowing errors in Fuyu is unrelated to the Trainer import issue caused by datasets." }, { - "left": "issue:44534", - "right": "issue:44913", + "left": "issue:44610", + "right": "issue:46369", "accept": false, - "reason": "One is a v5 buffer initialization regression; the other is a config value not persisting on reload." + "reason": "Processor image-size mismatch and SAM3 LiteText memory overflow are different bugs." }, { "left": "issue:45381", "right": "issue:47217", "accept": false, - "reason": "Both involve VL models, but one is about video position ids and the other about image placeholders in user text." + "reason": "Both are VL bugs, but one is about video position ids and the other is placeholder placement in chat text." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Serialization buffer junk and GPTNeoX rotary_pct reload drift are separate config/state issues." }, { "left": "issue:44792", "right": "issue:45600", "accept": false, - "reason": "A failed Janus test and a stale auto_mappings reference are separate issues." + "reason": "A Janus test failure is unrelated to stale auto_mappings references breaking CI." }, { "left": "issue:45542", "right": "issue:46234", "accept": false, - "reason": "Different backend/import warnings affecting different optional dependencies." + "reason": "TensorBoard/TF backend handling and missing falcon-mamba availability helper are unrelated." }, { "left": "issue:27957", "right": "issue:29925", "accept": false, - "reason": "XLMRoberta FlashAttention support and SigLIP fast-tokenizer enforcement are unrelated." + "reason": "XLMRoberta FlashAttention2 support and SigLIP use_fast enforcement are different model/processor issues." }, { "left": "issue:43525", "right": "issue:43531", "accept": false, - "reason": "Llama4Config pad_token_id handling and Qwen3-MoE sliding_window behavior are different bugs." + "reason": "Missing pad_token_id in Llama4Config and Qwen3-MoE sliding_window behavior are unrelated." }, { "left": "issue:43504", "right": "issue:45698", "accept": false, - "reason": "Legacy preset loading failure vs wrong custom module resolution after save/load are not the same code-path problem." + "reason": "Legacy-field preset loading and wrong custom module after save_pretrained are different loading bugs." }, { - "left": "issue:23647", - "right": "issue:27957", + "left": "issue:35141", + "right": "issue:42490", "accept": false, - "reason": "Wandb sweeps custom objective and XLMRoberta FlashAttention are unrelated." + "reason": "Embedding reinitialization on resize and trailing-slash local path warnings do not share the same root cause." }, { - "left": "issue:35141", - "right": "issue:42490", + "left": "issue:23647", + "right": "issue:27957", "accept": false, - "reason": "Embedding reinitialization after resize and trailing-slash local path warnings are unrelated." + "reason": "Wandb sweeps custom objective integration is unrelated to FlashAttention2 support." }, { "left": "issue:46693", "right": "issue:47217", "accept": false, - "reason": "Performance regression with FA2 and long-context generation is unrelated to image-placeholder placement failures." + "reason": "Long-context FA2 performance regression and VL placeholder parsing are different failures." }, { "left": "issue:45803", "right": "issue:46170", "accept": false, - "reason": "Bare except in tensor conversion and Windows/ROCm import errors are separate problems." + "reason": "One is exception handling in a converter; the other is a Windows/ROCm import dependency problem." }, { "left": "issue:43931", "right": "issue:44060", "accept": false, - "reason": "Qwen3-VL weight shape mismatch and Qwen3-Next tied-weights warning describe different model/config issues." + "reason": "Both mention weights, but one is a load-time shape mismatch and the other is a tied-weights warning bug." }, { "left": "issue:45632", "right": "issue:46424", "accept": false, - "reason": "Cache-path collisions for local models and max_cache_length handling are different cache bugs." - }, - { - "left": "issue:45588", - "right": "issue:47013", - "accept": false, - "reason": "FlashAttention sinkless-model crash and xLSTM state unpacking are unrelated failures." + "reason": "Cache-path collisions for trust_remote_code and max_cache_length behavior are unrelated cache bugs." }, { "left": "issue:44855", "right": "issue:45588", "accept": false, - "reason": "Python 3.13 IndentationError in DebertaV2 vs flash_attention.py AttributeError are unrelated." + "reason": "Python 3.13 decorator parsing and flash_attention s_aux=None crashes have no shared code path." }, { - "left": "issue:46082", - "right": "issue:46697", + "left": "issue:45588", + "right": "issue:47013", "accept": false, - "reason": "LlamaConfig head_dim validation and Swin2SR image over-padding are different subsystems." + "reason": "Both crash at runtime, but in different layers and with different causes." }, { "left": "issue:45127", "right": "issue:45200", "accept": false, - "reason": "LoRA merge/tied-embeddings collapse and Gemma 4 mm_token_type_ids defaults are not the same issue." + "reason": "LoRA merge collapse with tied embeddings and Gemma 4 mm_token_type_ids defaults are not the same defect." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "A Janus test failure is unrelated to an unreleased MIN_PEFT_VERSION bump." }, { "left": "issue:41950", "right": "issue:43475", "accept": false, - "reason": "Video-classification processor lookup and Sam3VisionEncoderOutput missing attribute are unrelated bugs." + "reason": "Video-classification processor selection and SAM3 VisionEncoderOutput missing a field are different bugs." } ] }, @@ -12478,7 +12421,6 @@ "issue:43065", "issue:43335", "issue:43388", - "issue:43526", "issue:43550", "issue:43644", "issue:43650", @@ -12502,13 +12444,14 @@ "issue:45072", "issue:45245", "issue:45381", - "issue:45405", "issue:45406", "issue:45563", "issue:45584", + "issue:46082", "issue:46202", "issue:46369", "issue:46399", + "issue:46655", "issue:46697", "issue:46832", "issue:46858", @@ -12519,17 +12462,17 @@ ], "soft_pairs": [ "issue:43673|issue:44743", - "issue:44792|issue:45405", "issue:44242|issue:46894", + "issue:46082|issue:46697", "issue:43650|issue:46202", - "issue:44617|issue:45584", "issue:44190|issue:44964", + "issue:44617|issue:45584", "issue:25251|issue:29870", "issue:43827|issue:43994", "issue:43388|issue:43688", - "issue:45245|issue:46858", - "issue:45072|issue:46932", "issue:29870|issue:35141", + "issue:45072|issue:46932", + "issue:45245|issue:46858", "issue:43644|issue:44913", "issue:44734|issue:45406", "issue:44869|issue:46832", @@ -12538,218 +12481,218 @@ "issue:23647|issue:29127", "issue:43854|issue:44792", "issue:45381|issue:46369", - "issue:41720|issue:45245", "issue:39692|issue:47321", - "issue:43550|issue:44190", "issue:43827|issue:47321", + "issue:41720|issue:45245", + "issue:43550|issue:44190", "issue:45563|issue:45584", "issue:44617|issue:46369", "issue:43065|issue:44610", "issue:44479|issue:44734", - "issue:43526|issue:46697" + "issue:45381|issue:46655" ], - "prepared_review_unit_hash": "19ffb11eb8f983f018f53b397e19706efd2663e279eb0fcfa2b7dc915b83dc39", + "prepared_review_unit_hash": "710280380da3cd1b73881f1232c60ee086c9f74ebc78e5bd62cf8574232e5532", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36520, - "estimated_input_tokens": 9130, - "estimated_eval_tokens": 18516 + "serialized_chars": 36514, + "estimated_input_tokens": 9129, + "estimated_eval_tokens": 18514 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the issues span unrelated models, docs, serve/runtime, quantization, generation, and data-processing bugs. I do not see any soft pair that clearly describes the same underlying bug or change, so none should be merged.", - "confidence": 0.94, - "canonical_issue_reason": "issue:43644 is the broadest, most cross-cutting regression in the set and is the best anchor for the cluster; the rest are mostly isolated model- or feature-specific bugs.", + "summary": "This cluster is highly heterogeneous: the soft pairs share only broad topical similarity, not the same underlying bug or change. I found no true duplicate pair, so the cluster should not be merged on these edges.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43644 is the broadest, most cross-cutting regression in the set (framework-level buffer initialization behavior) and is more representative of the cluster than the many narrowly scoped model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the strongest standalone issue here because it describes a systemic runtime problem rather than a narrow model-specific symptom.", + "best_issue_reason": "issue:43644 is the best single representative issue because it has the widest impact and clearest general runtime-regression framing among otherwise unrelated reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43673", "right": "issue:44743", "accept": false, - "reason": "Both involve cache/state behavior, but they target different code paths and models; not the same bug." + "reason": "Different failures: a generic GenerationMixin cache/chunked_prefill regression vs Qwen3.5 recurrent states resetting with cache and seq_len>1." }, { - "left": "issue:44792", - "right": "issue:45405", + "left": "issue:44242", + "right": "issue:46894", "accept": false, - "reason": "Completely unrelated: a Janus test failure versus a PEFT version bump/regression." + "reason": "Different model/loss bugs: MoE router load-balancing loss omission vs CohereASR double-shift training-loss bug." }, { - "left": "issue:44242", - "right": "issue:46894", + "left": "issue:46082", + "right": "issue:46697", "accept": false, - "reason": "Both mention loss handling, but the actual bugs are different training-loss issues in different models." + "reason": "Unrelated domains: LlamaConfig head_dim validation vs Swin2SR image padding in an image processor." }, { "left": "issue:43650", "right": "issue:46202", "accept": false, - "reason": "No meaningful overlap; one is a vague placeholder issue, the other is a Gemma pipeline roundtrip bug." + "reason": "No shared bug shape: a vague data request vs a Gemma pipeline roundtrip/tokenization issue." }, { - "left": "issue:44617", - "right": "issue:45584", + "left": "issue:44190", + "right": "issue:44964", "accept": false, - "reason": "OOM in Sam3Video and Whisper empty-transcription failure are unrelated failures." + "reason": "Different load paths and symptoms: local dataset loading in a training script vs loading a multimodal model." }, { - "left": "issue:44190", - "right": "issue:44964", + "left": "issue:44617", + "right": "issue:45584", "accept": false, - "reason": "A local dataset loading problem and a multimodal model loading issue are distinct." + "reason": "Different subsystems: Sam3Video OOM vs Whisper empty-transcription failure after align_special_tokens." }, { "left": "issue:25251", "right": "issue:29870", "accept": false, - "reason": "Different subsystems and failure modes: pipeline output shape versus BART generate masking." + "reason": "Different pipeline/eval bugs: top_k nested-list output vs BART generate() unbound local error." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "Docs referencing removed pipeline APIs is not the same as SigLIP2 producing wrong results." + "reason": "Docs referencing removed pipeline() APIs is not the same as SigLIP2 inference producing nonsensical results." }, { "left": "issue:43388", "right": "issue:43688", "accept": false, - "reason": "Different bugs in different training/evaluation paths; only a vague loss-related similarity." + "reason": "Different training/eval bugs: gather_for_metrics label truncation vs auxiliary-loss normalization." }, { - "left": "issue:45245", - "right": "issue:46858", + "left": "issue:29870", + "right": "issue:35141", "accept": false, - "reason": "One is an embedding/category limit error; the other is a cache length tensor type issue." + "reason": "Different code paths: BART generate() assignment bug vs embedding resize/post_init weight reinitialization." }, { "left": "issue:45072", "right": "issue:46932", "accept": false, - "reason": "Both are dtype-related, but they affect different models and different quantization/inference code paths." + "reason": "Both mention dtype issues, but they affect different models and mechanisms; not the same concrete bug." }, { - "left": "issue:29870", - "right": "issue:35141", + "left": "issue:45245", + "right": "issue:46858", "accept": false, - "reason": "Generate-time decoder mask error versus post-init embedding reinitialization are unrelated." + "reason": "Category-count limit error is unrelated to the static-cache slice arithmetic/device-tensor bug." }, { "left": "issue:43644", "right": "issue:44913", "accept": false, - "reason": "A general buffer initialization regression is not the same as GPTNeoX config reload losing rotary_pct." + "reason": "Different regressions: non-persistent buffer initialization junk vs GPTNeoXConfig reload dropping rotary_pct." }, { "left": "issue:44734", "right": "issue:45406", "accept": false, - "reason": "Both touch serve, but one is KV-cache continuation indexing and the other is a processor attribute error." + "reason": "Both involve transformers serve, but one is KV-cache continuation indexing and the other is a missing _tokenizer attribute." }, { "left": "issue:44869", "right": "issue:46832", "accept": false, - "reason": "Whisper token decoding crash and RT-DETR feature-level mismatch are unrelated." + "reason": "Whisper word-timestamp decoding crash is unrelated to RT-DETR feature-level/backbone mismatch." }, { "left": "issue:43335", "right": "issue:44855", "accept": false, - "reason": "Sparse-layer config bug and Python 3.13 decorator parsing failure have no common code path." + "reason": "Sparse-layer creation in SwitchTransformersConfig is unrelated to the Python 3.13 IndentationError/JIT parsing issue." }, { "left": "issue:46399", "right": "issue:47312", "accept": false, - "reason": "Both mention kernels, but one is a specific registration crash and the other is a broad compatibility complaint." + "reason": "Kernel compatibility concerns overlap only loosely; one is a missing decorator registration bug in Qwen3.5, the other is a general kernels version incompatibility report." }, { "left": "issue:23647", "right": "issue:29127", "accept": false, - "reason": "WandB sweep objective handling is unrelated to LayoutLMv3 error messaging." + "reason": "Wandb sweep custom objective integration is unrelated to LayoutLMv3 box-information error handling." }, { "left": "issue:43854", "right": "issue:44792", "accept": false, - "reason": "Different models and failures: GLM-4.7-Flash loading versus Janus image-generation test failure." + "reason": "GLM-4.7-Flash loading problems are not the same as a janus image-generation test failure." }, { "left": "issue:45381", "right": "issue:46369", "accept": false, - "reason": "Both are model-specific regressions, but they concern different models and symptoms." + "reason": "Qwen2.5-VL video position IDs bug is unrelated to Sam3-litetext memory overflow." }, { - "left": "issue:41720", - "right": "issue:45245", + "left": "issue:39692", + "right": "issue:47321", "accept": false, - "reason": "Auto device mapping CUDA assert and category-count overflow are unrelated." + "reason": "SigLIP2 doc/example issues and AMD quark class updates are unrelated." }, { - "left": "issue:39692", + "left": "issue:43827", "right": "issue:47321", "accept": false, - "reason": "SigLIP2 docs/model mismatch and AMD quark class update are unrelated." + "reason": "Docs still using pipeline() and AMD quark class updates are unrelated." }, { - "left": "issue:43550", - "right": "issue:44190", + "left": "issue:41720", + "right": "issue:45245", "accept": false, - "reason": "torch.compile/SDPA failure is unrelated to local dataset loading in an example script." + "reason": "A800 cudaErrorAssert in Qwen3 auto device mapping is unrelated to the 2^24 category limit error." }, { - "left": "issue:43827", - "right": "issue:47321", + "left": "issue:43550", + "right": "issue:44190", "accept": false, - "reason": "Docs cleanup and AMD backend class update are unrelated." + "reason": "torch.compile/SDPA failure in Bamba is unrelated to local dataset loading in a training script." }, { "left": "issue:45563", "right": "issue:45584", "accept": false, - "reason": "A stale warning in paged generate is not the same as Whisper failing on empty transcription." + "reason": "Stale warning for num_return_sequences is unrelated to Whisper empty-transcription failure." }, { "left": "issue:44617", "right": "issue:46369", "accept": false, - "reason": "Both mention memory pressure in Sam3, but one is video OOM and the other is a different submodel's overflow symptom." + "reason": "Both are memory-related, but they are different models and different failure modes (OOM vs memory overflow)." }, { "left": "issue:43065", "right": "issue:44610", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and OmDet-Turbo input-size mismatch are different model/input issues." + "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same bug as OmDet-Turbo processor/model input-size mismatch." }, { "left": "issue:44479", "right": "issue:44734", "accept": false, - "reason": "Video-input regression in Qwen VL models and serve KV-cache indexing crash are different bugs." + "reason": "Video-input regression across Qwen models is unrelated to the serve KV-cache indexing crash." }, { - "left": "issue:43526", - "right": "issue:46697", + "left": "issue:45381", + "right": "issue:46655", "accept": false, - "reason": "Label reduction bug and image over-padding bug are unrelated preprocessing issues." + "reason": "Both involve video processors, but one is Qwen2.5-VL vision_position_ids and the other is SmolVLM padding calculation." } ] }, @@ -12774,13 +12717,13 @@ "cluster_id": "cluster-10000-540", "nodes": [ "issue:27957", - "issue:32090", "issue:35141", "issue:39692", "issue:42175", "issue:43295", "issue:43317", "issue:43335", + "issue:43526", "issue:43540", "issue:43611", "issue:43643", @@ -12804,6 +12747,7 @@ "issue:45276", "issue:45357", "issue:45362", + "issue:45375", "issue:45381", "issue:45520", "issue:45584", @@ -12814,7 +12758,7 @@ "issue:46421", "issue:46489", "issue:46617", - "issue:46655", + "issue:46697", "issue:46858", "issue:46897", "issue:47145", @@ -12822,245 +12766,241 @@ "issue:47321" ], "soft_pairs": [ - "issue:45381|issue:46655", + "issue:43526|issue:46697", "issue:45632|issue:46617", "issue:43295|issue:45276", "issue:43335|issue:44190", + "issue:43643|issue:45375", "issue:39692|issue:44190", - "issue:44779|issue:46489", "issue:44933|issue:45362", + "issue:44779|issue:46489", "issue:43854|issue:44964", "issue:43540|issue:44734", "issue:45276|issue:45584", "issue:43994|issue:47321", "issue:47145|issue:47321", - "issue:43673|issue:46421", - "issue:43643|issue:44496", "issue:44242|issue:46897", + "issue:43643|issue:44496", + "issue:43673|issue:46421", "issue:45127|issue:45357", "issue:45520|issue:46170", - "issue:44521|issue:45381", "issue:43611|issue:47321", + "issue:44521|issue:45381", "issue:43317|issue:47312", "issue:44112|issue:45600", "issue:42175|issue:44279", "issue:44112|issue:44964", "issue:44246|issue:45003", - "issue:44462|issue:45907", "issue:27957|issue:35141", - "issue:43673|issue:46858", - "issue:44964|issue:45600", - "issue:32090|issue:35141" + "issue:44462|issue:45907", + "issue:43673|issue:46858" ], - "prepared_review_unit_hash": "8a571a1dbc9dd10159e660f67bd24bcf8bd1f6cf7910ec096bf2accd1160185b", + "prepared_review_unit_hash": "fbf0671c2eb8d8919c39958331be706cdf89d5a7e39730354e91fd13b3b0c619", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35854, - "estimated_input_tokens": 8964, - "estimated_eval_tokens": 18184 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36171, + "estimated_input_tokens": 9043, + "estimated_eval_tokens": 18342 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the issues span unrelated areas like multimodal/video processing, tokenizer regressions, cache/chunked-prefill behavior, model loading, packaging, and CI/test failures. I do not see any clear duplicate pair among the proposed soft edges.", - "confidence": 0.96, - "canonical_issue_reason": "No canonical issue: the items span many unrelated failure modes with no single underlying bug.", + "summary": "This cluster is largely heterogeneous: it mixes unrelated bugs in tokenizers, multimodal processors, cache handling, configs, and CI. Only two soft pairs look like the same underlying issue: the DeepSeek tokenizer regression and the trust_remote_code cache-loading failures.", + "confidence": 0.57, + "canonical_issue_reason": "No single coherent duplicate theme spans the whole set, so the best representative is the most active open issue, issue 44521, which has the strongest discussion/inbound signal and is broadly scoped around multimodal chat-template behavior.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a strong cluster representative for the same reason: the cluster is not centered on one bug or code path.", + "best_issue_reason": "Issue 44521 is the most suitable representative for the cluster overall because it is open, has the most engagement, and is the broadest live bug report among the items.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45381", - "right": "issue:46655", + "left": "issue:43526", + "right": "issue:46697", "accept": false, - "reason": "Both involve video/multimodal inputs, but one is Qwen2.5-VL vision_position_ids and the other is SmolVLM video padding; different code paths and symptoms." + "reason": "Different code paths: BeitImageProcessorFast label reduction vs Swin2SR image padding. No shared underlying bug." }, { "left": "issue:45632", "right": "issue:46617", - "accept": false, - "reason": "Both touch trust_remote_code/cache behavior, but one is a leaf-directory cache collision for local models and the other is a FileNotFoundError when loading from cache." + "accept": true, + "reason": "Both point to trust_remote_code cache/path resolution failures when loading models from cache; likely the same cache-path bug manifesting differently." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "Both mention model-loading/output behavior after API changes, but one is a processor.tokenizer regression and the other is resize_token_embeddings not updating Gemma4 outputs." + "reason": "Different regression areas: processor/tokenizer access in custom multimodal code vs Gemma4 resize_token_embeddings not updating embeddings." }, { "left": "issue:43335", "right": "issue:44190", "accept": false, - "reason": "Sparse-layer construction in SwitchTransformers is unrelated to loading a local dataset in a training script." + "reason": "SwitchTransformers sparse-layer config bug and local dataset loading failure in an example script are unrelated." }, { - "left": "issue:39692", - "right": "issue:44190", + "left": "issue:43643", + "right": "issue:45375", "accept": false, - "reason": "SigLIP2 doc-example errors are unrelated to the local-dataset loading failure in the no-trainer image classification script." + "reason": "One is a trust_remote_code AutoConfig field-loss issue; the other is a specific config field being dropped by @strict. Similar theme, not the same bug." }, { - "left": "issue:44779", - "right": "issue:46489", + "left": "issue:39692", + "right": "issue:44190", "accept": false, - "reason": "Both are DeepSeek tokenizer complaints, but one is a general v5 regression and the other is specifically DeepSeek-Coder v1 with a fix gap; not clearly the same bug." + "reason": "SigLIP2 doc example errors are unrelated to local dataset loading in run_image_classification_no_trainer.py." }, { "left": "issue:44933", "right": "issue:45362", "accept": false, - "reason": "A missing import from image_utils is unrelated to Qwen3.5 chat crashes." + "reason": "Nonexistent image_utils import and Qwen3.5 chat crashes are different failures." + }, + { + "left": "issue:44779", + "right": "issue:46489", + "accept": true, + "reason": "Both describe DeepSeek tokenizer regressions in v5; the later issue explicitly frames it as a gap in the prior fix, so they appear to be the same underlying tokenizer bug." }, { "left": "issue:43854", "right": "issue:44964", "accept": false, - "reason": "Both are model-loading failures, but for different models and different failure modes." + "reason": "Both are model-loading complaints, but they involve different models and likely different compatibility failures." }, { "left": "issue:43540", "right": "issue:44734", "accept": false, - "reason": "Qwen3OmniMoe video-input validation and transformers serve KV-cache continuation crashes are different problems." + "reason": "Video-input ValueError in Qwen3OmniMoe and KV-cache continuation crash in transformers serve are different code paths." }, { "left": "issue:45276", "right": "issue:45584", "accept": false, - "reason": "Resize-token-embeddings/output-embedding updates are unrelated to Whisper empty-transcription generation failure." + "reason": "Embedding resize/update bug vs Whisper empty-transcription generation bug after align_special_tokens are unrelated." }, { "left": "issue:43994", "right": "issue:47321", "accept": false, - "reason": "SigLIP2 bad outputs and an AMD quark-class update issue are unrelated." + "reason": "SigLIP2 inference quality issue and AMD quark class update issue are unrelated." }, { "left": "issue:47145", "right": "issue:47321", "accept": false, - "reason": "FP8 weight_scale loading corruption is unrelated to the AMD quark class update issue." + "reason": "FP8 weight_scale loading corruption is unrelated to AMD quark class updates." }, { - "left": "issue:43673", - "right": "issue:46421", + "left": "issue:44242", + "right": "issue:46897", "accept": false, - "reason": "Both involve chunked prefill/cache, but one is a missing cache in v5 and the other is recompilation from lazy cache initialization; different concrete bugs." + "reason": "Load-balancing loss omission in MoE routing and Florence2 double-shift training loss are different bugs." }, { "left": "issue:43643", "right": "issue:44496", "accept": false, - "reason": "Missing fields in AutoConfig with trust_remote_code and an unrecognized-model/model_type error are distinct config-loading issues." + "reason": "trust_remote_code field loss and unrecognized model_type in config.json are different loading/configuration problems." }, { - "left": "issue:44242", - "right": "issue:46897", + "left": "issue:43673", + "right": "issue:46421", "accept": false, - "reason": "Load-balancing loss omission and Florence2 double-shift training loss are different loss-calculation bugs." + "reason": "Both mention chunked prefill/cache, but one is a missing GenerationMixin cache and the other is lazy-init recompilation; not the same concrete bug." }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "LoRA/extended-vocabulary collapse with tied embeddings is unrelated to incorrect visual encoder keys in save_pretrained." + "reason": "LoRA/extended-vocab collapse and incorrect visual encoder keys in save_pretrained are unrelated." }, { "left": "issue:45520", "right": "issue:46170", "accept": false, - "reason": "A Python 3.13 flash_attn import bug and a Windows ROCm/torch.distributed dependency issue are unrelated platform/import failures." + "reason": "flash_attn import on Python 3.13 and Windows ROCm torch.distributed dependency issues are unrelated." }, { - "left": "issue:44521", - "right": "issue:45381", + "left": "issue:43611", + "right": "issue:47321", "accept": false, - "reason": "Both are multimodal, but apply_chat_template assistant-mask generation and Qwen2.5-VL video vision_position_ids are different code paths." + "reason": "base_model_prefix loading regression and AMD quark class update are unrelated." }, { - "left": "issue:43611", - "right": "issue:47321", + "left": "issue:44521", + "right": "issue:45381", "accept": false, - "reason": "Base_model_prefix loading regression in Transformers 5.0.0 is unrelated to the AMD quark class issue." + "reason": "Both are multimodal, but assistant_masks being all zero is a different bug from wrong vision_position_ids in qwen2.5-vl video inputs." }, { "left": "issue:43317", "right": "issue:47312", "accept": false, - "reason": "Dequantized-model offload loading and kernel-version incompatibility are unrelated." + "reason": "dequantized model offload failure and kernel compatibility issue are unrelated." }, { "left": "issue:44112", "right": "issue:45600", "accept": false, - "reason": "A stale CI test and outdated auto_mappings references to removed configs are different failures." + "reason": "A stale GraniteSpeech CI test and removed Sam3LiteText auto-mappings are different CI failures." }, { "left": "issue:42175", "right": "issue:44279", "accept": false, - "reason": "A missing TensorFlow backend in the torch extra and a general dependency issue are too broad and not the same concrete bug." + "reason": "TensorFlow backend packaging and a generic dependency issue are not the same bug." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "GraniteSpeech CI test flakiness and Phi-4 multimodal loading are unrelated." + "reason": "GraniteSpeech CI test failure is unrelated to Phi-4 multimodal model loading." }, { "left": "issue:44246", "right": "issue:45003", "accept": false, - "reason": "Import latency and unsafe sys.modules access are related only at a high level; not the same bug or fix." - }, - { - "left": "issue:44462", - "right": "issue:45907", - "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are different repository-resolution problems." + "reason": "Import slowdown and unsafe sys.modules access are related at a high level, but not the same concrete failure." }, { "left": "issue:27957", "right": "issue:35141", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta is unrelated to embedding reinitialization after resize_token_embeddings." - }, - { - "left": "issue:43673", - "right": "issue:46858", - "accept": false, - "reason": "Both involve cache/chunked prefill, but one is a missing cache attribute and the other is a device-tensor seq_length arithmetic bug." + "reason": "XLMRoberta Flash Attention support and post_init reinitializing output embeddings after resize are unrelated." }, { - "left": "issue:44964", - "right": "issue:45600", + "left": "issue:44462", + "right": "issue:45907", "accept": false, - "reason": "Phi-4 multimodal loading and stale auto_mappings references to removed configs are different issues." + "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are different API issues." }, { - "left": "issue:32090", - "right": "issue:35141", + "left": "issue:43673", + "right": "issue:46858", "accept": false, - "reason": "Trainer _gpu_broadcast_one TypeError is unrelated to embedding reinitialization after resizing." + "reason": "Both concern cache/prefill, but one is a missing cache in v5.0.0 and the other is a device-tensor seq_length causing slice arithmetic failures; not the same bug." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The cluster is correctly characterized as heterogeneous, the representative issue choice is supported by activity/open-state signals, and the two accepted soft pairs are at least plausibly same-bug matches based on the packet titles. No clear overstatement of evidence." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -13079,7 +13019,9 @@ "nodes": [ "issue:25251", "issue:29925", + "issue:32090", "issue:33290", + "issue:35141", "issue:36246", "issue:38617", "issue:39692", @@ -13089,20 +13031,19 @@ "issue:42994", "issue:43295", "issue:43335", - "issue:43381", "issue:43404", "issue:43531", "issue:43673", "issue:43976", "issue:44016", "issue:44279", - "issue:44295", "issue:44336", "issue:44496", "issue:44521", "issue:44683", "issue:44737", "issue:44938", + "issue:44964", "issue:44987", "issue:45042", "issue:45083", @@ -13126,6 +13067,8 @@ "issue:47333" ], "soft_pairs": [ + "issue:32090|issue:35141", + "issue:44964|issue:45600", "issue:43404|issue:45216", "issue:42796|issue:44521", "issue:43531|issue:45600", @@ -13135,215 +13078,213 @@ "issue:45290|issue:46752", "issue:45632|issue:46234", "issue:44496|issue:45042", - "issue:44938|issue:45542", "issue:36246|issue:38617", "issue:25251|issue:29925", "issue:36246|issue:41628", + "issue:44938|issue:45542", "issue:43295|issue:45335", "issue:45083|issue:46506", "issue:44496|issue:45759", "issue:41950|issue:44016", - "issue:43976|issue:44987", "issue:29925|issue:33290", + "issue:43976|issue:44987", "issue:39692|issue:46932", "issue:44683|issue:45446", - "issue:44295|issue:46506", - "issue:43673|issue:46424", - "issue:43673|issue:44336", - "issue:44279|issue:46515", "issue:42994|issue:47333", - "issue:43381|issue:45593" + "issue:44279|issue:46515", + "issue:43673|issue:46424", + "issue:43673|issue:44336" ], - "prepared_review_unit_hash": "29cc84d411ba23f92f31b13793b4be09d1a4d0b960a8fae0bba531208d73ad75", + "prepared_review_unit_hash": "e053a2b815bd1b506a9d3fb8fde16800591d1c9eab20ffa97ef738e42c3f9761", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35193, - "estimated_input_tokens": 8799, - "estimated_eval_tokens": 17854 + "serialized_chars": 35866, + "estimated_input_tokens": 8967, + "estimated_eval_tokens": 18190 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These soft pairs are mostly not duplicates: they cover different bugs across model loading, tokenizers, chat templates, configs, cache behavior, flex attention, and docs. One pair is thematically close (two `apply_chat_template` crashes), but the triggers and failure modes are different, so I would not merge it.", - "confidence": 0.95, - "canonical_issue_reason": "No single canonical issue fits this set: the items are heterogeneous and do not form one duplicate bug cluster.", + "summary": "This cluster is mostly a semantic false-positive around assorted Transformers bugs. None of the proposed pairs look like the same concrete bug/change, so I would not merge any soft edge.", + "confidence": 0.69, + "canonical_issue_reason": "Issue #44496 is the most representative anchor in this noisy set: it is a concrete model-loading/configuration failure and has the strongest overlap with the nearby import/load-related reports, though the cluster is too heterogeneous for true deduping.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue is suitable as an umbrella; the closest item is `issue:44521`, but it only covers one multimodal chat-template bug and does not represent the rest of the set.", + "best_issue_reason": "Issue #44496 is the clearest single user-facing bug to anchor triage in this cluster, but it should be treated as a representative example rather than a duplicate sink for the other issues.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Different failures and code paths: GPU broadcast/type handling in Trainer vs embedding resize/post-init reinitialization." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Model-loading failure for a specific checkpoint vs CI break from stale auto_mappings references; not the same bug." + }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "Different regressions in different code paths: Mistral3 weight tying vs Qwen3.5 checkpoint saving." + "reason": "Both involve model saving/weight tying, but they are different models and different failure modes, so not one concrete bug." }, { "left": "issue:42796", "right": "issue:44521", "accept": false, - "reason": "Tokenizer BPE marker handling and multimodal assistant mask generation are unrelated bugs." + "reason": "Tokenizer BPE-marker handling is unrelated to the multimodal assistant-mask bug in apply_chat_template." }, { "left": "issue:43531", "right": "issue:45600", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to stale auto-mappings references for Sam3LiteText configs." + "reason": "Qwen3-MoE sliding-window behavior and removed-config CI failures are unrelated." }, { "left": "issue:44737", "right": "issue:45803", "accept": false, - "reason": "CPU device placement in XLNet positional encoding is a different bug from Fuyu's exception handling." + "reason": "CPU device-placement bug in XLNet vs bare-except error masking in Fuyu batch conversion; different code paths." }, { "left": "issue:44496", "right": "issue:47217", "accept": false, - "reason": "Model-type config loading failure is unrelated to image placeholders inside user text for VL models." + "reason": "Model/config loading error vs image-placeholder parsing failure in VL text; not the same underlying issue." }, { "left": "issue:43335", "right": "issue:45593", "accept": false, - "reason": "SwitchTransformers sparse-layer construction and D-FINE auxiliary-loss logic are separate model bugs." + "reason": "Sparse-layer construction bug vs auxiliary-loss gating in D-FINE; distinct behavior and fix." }, { "left": "issue:45290", "right": "issue:46752", "accept": false, - "reason": "Both involve `apply_chat_template`, but one is tool-call assistant content and the other is an empty list; different inputs and likely different failure paths." + "reason": "Both crash in apply_chat_template, but one is about tool-call assistant messages and the other is an empty-list IndexError; not clearly the same bug." }, { "left": "issue:45632", "right": "issue:46234", "accept": false, - "reason": "Cache-path collisions for local trust-remote-code models are unrelated to a missing import utility symbol." + "reason": "Cache-path collision for local trust_remote_code models is unrelated to a missing import_utils helper." }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Unrecognized model/config loading is unrelated to PIL image processors requiring torchvision." - }, - { - "left": "issue:44938", - "right": "issue:45542", - "accept": false, - "reason": "Python 3.14 import/load failure is unrelated to the tensorboard-without-TensorFlow backend issue." + "reason": "Unrecognized model/config issue vs PIL image processors incorrectly requiring torchvision; different failures." }, { "left": "issue:36246", "right": "issue:38617", "accept": false, - "reason": "Different missing imports in different modules: Qwen2.5 VL image processor vs `layer_type_validation`." + "reason": "Different missing imports from different modules/symbols; same general symptom, not the same bug." }, { "left": "issue:25251", "right": "issue:29925", "accept": false, - "reason": "Pipeline `top_k` output-shape change and SigLIP fast-tokenizer enforcement are unrelated." + "reason": "Pipeline top_k nesting bug and SigLIP use_fast enforcement are unrelated." }, { "left": "issue:36246", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but for different symbols and modules with different root causes." + "reason": "Both are import errors, but for different symbols and different model families; too broad to merge." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python-version load failure vs TensorFlow/tensorboard backend misdetection; unrelated." }, { "left": "issue:43295", "right": "issue:45335", "accept": false, - "reason": "Processor/tokenizer regression and T5Gemma embedding resizing are different code paths and bugs." + "reason": "Processor/tokenizer regression and resize_token_embeddings not propagating to decoder.embed_tokens are separate bugs." }, { "left": "issue:45083", "right": "issue:46506", "accept": false, - "reason": "Qwen3 Omni-MoE feature-length helper behavior and PPDocLayoutV3 CPU/CUDA divergence are unrelated." + "reason": "Helper function length calculation in qwen3_omni_moe vs CPU/CUDA divergence in a different model; not the same bug." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "Config recognition failure and `text_config` unwrapping in multimodal `from_config` are different issues." + "reason": "Generic model registration/config failure vs composite multimodal AutoModelForCausalLM unwrapping bug; different code paths." }, { "left": "issue:41950", "right": "issue:44016", "accept": false, - "reason": "Pipeline processor lookup bug and notebook syntax error are not the same underlying problem." + "reason": "Video-classification processor lookup and a notebook syntax error are unrelated." }, { - "left": "issue:43976", - "right": "issue:44987", + "left": "issue:29925", + "right": "issue:33290", "accept": false, - "reason": "General Python version support failure and loading one specific model on 5.1.0 are different bugs." + "reason": "Tokenizer fast-flag enforcement and Deepspeed/Adafactor OOM are not the same issue." }, { - "left": "issue:29925", - "right": "issue:33290", + "left": "issue:43976", + "right": "issue:44987", "accept": false, - "reason": "Tokenizer fast-path enforcement is unrelated to an Adafactor/DeepSpeed OOM." + "reason": "Python 3.9/3.10 compatibility problem vs a specific model-loading failure; not the same change or bug." }, { "left": "issue:39692", "right": "issue:46932", "accept": false, - "reason": "SigLIP2 doc example errors and Gemma4 quantized embedder dtype casting are separate concerns." + "reason": "SigLIP2 documentation errors and Gemma4 quantized dtype-casting are different model-specific issues." }, { "left": "issue:44683", "right": "issue:45446", "accept": false, - "reason": "Compiled flex_attention failing on torch>=2.9 is different from a bad PyTorch version check for AuxRequest import." - }, - { - "left": "issue:44295", - "right": "issue:46506", - "accept": false, - "reason": "Position_ids buffer access and CPU/CUDA output divergence are unrelated bugs." - }, - { - "left": "issue:43673", - "right": "issue:46424", - "accept": false, - "reason": "Chunked-prefill cache missing in GenerationMixin and cache max length handling for quantized cache are different cache features." + "reason": "Compiled flex_attention failing on torch >= 2.9 is distinct from the separate PyTorch version check bug for AuxRequest import." }, { - "left": "issue:43673", - "right": "issue:44336", + "left": "issue:42994", + "right": "issue:47333", "accept": false, - "reason": "Generation cache behavior and ANSI-color reporting are unrelated." + "reason": "Quantized model saving and disk-offloaded model saving are related only at a high level; different failure modes and fixes." }, { "left": "issue:44279", "right": "issue:46515", "accept": false, - "reason": "A generic dependency issue and a torch 2.4.1 incompatibility are too broad and not the same concrete failure." + "reason": "Generic dependency issue vs torch 2.4.1 incompatibility; too broad and not the same concrete bug." }, { - "left": "issue:42994", - "right": "issue:47333", + "left": "issue:43673", + "right": "issue:46424", "accept": false, - "reason": "Quantized model saving and disk-offloaded model saving are different save paths with different bugs." + "reason": "GenerationMixin cache missing in chunked_prefill is separate from max_cache_length only applying to Quantized Cache." }, { - "left": "issue:43381", - "right": "issue:45593", + "left": "issue:43673", + "right": "issue:44336", "accept": false, - "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary losses are unrelated model behaviors." + "reason": "Cache/generation behavior vs ANSI formatting in loading_report; unrelated." } ] }, @@ -13374,6 +13315,7 @@ "issue:42994", "issue:43295", "issue:43299", + "issue:43381", "issue:43441", "issue:43479", "issue:43526", @@ -13387,6 +13329,7 @@ "issue:44062", "issue:44220", "issue:44279", + "issue:44295", "issue:44327", "issue:44466", "issue:44496", @@ -13409,20 +13352,21 @@ "issue:46123", "issue:46369", "issue:46424", - "issue:46614", + "issue:46506", "issue:46752", - "issue:47085", "issue:47217", "issue:47332" ], "soft_pairs": [ + "issue:44295|issue:46506", + "issue:43381|issue:45593", "issue:43526|issue:43901", "issue:44327|issue:46752", "issue:44016|issue:44220", "issue:46082|issue:46121", "issue:42994|issue:47332", - "issue:44016|issue:44062", "issue:43650|issue:44220", + "issue:44016|issue:44062", "issue:43441|issue:45588", "issue:44792|issue:46424", "issue:42907|issue:43441", @@ -13433,8 +13377,8 @@ "issue:43901|issue:44496", "issue:43976|issue:44279", "issue:43295|issue:44779", - "issue:43901|issue:47217", "issue:39692|issue:43901", + "issue:43901|issue:47217", "issue:44779|issue:45812", "issue:43299|issue:45216", "issue:42796|issue:43650", @@ -13443,211 +13387,216 @@ "issue:44466|issue:45276", "issue:43577|issue:45216", "issue:46123|issue:46369", - "issue:46614|issue:47085", "issue:43653|issue:45593" ], - "prepared_review_unit_hash": "e8ede60f3b3a0ca3297fa999ef70bd220006286abfd69bbe567d47d18d3b7815", + "prepared_review_unit_hash": "23a28430c7f36e36dfb872e240c9e3f50636abd46ca0026f2118f479053f9355", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35852, - "estimated_input_tokens": 8963, - "estimated_eval_tokens": 18182 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36672, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is highly heterogeneous: most items are unrelated bug reports spanning tokenizers, saving/loading, configs, pipelines, and docs. None of the soft pairs appear to be the same underlying issue, so no duplicates should be merged.", - "confidence": 0.92, - "canonical_issue_reason": "No single issue is a clear canonical here; the candidate pairs only overlap at a broad subsystem level and describe different concrete failures.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers regressions, docs issues, tokenizer/model bugs, and save/load edge cases. I did not find any soft pair that looks like the same underlying bug/change, so no duplicate merges are justified.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 45020 is the broadest and most central regression report in the set, but the items here are not true duplicates; it is only the best umbrella-style representative if one must be chosen.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong global representative because the collection is not one cohesive bug cluster.", + "best_issue_reason": "Issue 45020 is the strongest standalone issue to keep because it covers a wide, user-impacting regression class (recent versions breaking remote-code models) and is more general than the others.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44295", + "right": "issue:46506", + "accept": false, + "reason": "Different bugs: one is a position_ids buffer-read error, the other is a CPU/CUDA output divergence in PPDocLayoutV3." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Different code paths and symptoms: eval/gradient-checkpointing conflict vs D-FINE auxiliary-loss handling when denoising is off." + }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "Different problems: BeitImageProcessorFast label reduction vs a docs/behavior mismatch for TextClassificationPipeline." + "reason": "One is a BeitImageProcessorFast label-reduction bug; the other is a docs/behavior mismatch for TextClassificationPipeline." }, { "left": "issue:44327", "right": "issue:46752", "accept": false, - "reason": "QA decode_spans top-k bounds crash and empty chat-template input crash are unrelated code paths." + "reason": "Different pipelines and failures: QA decode_spans top-k bounds error vs empty-list apply_chat_template crash." }, { "left": "issue:44016", "right": "issue:44220", "accept": false, - "reason": "Notebook syntax error vs audio feature extraction TypeError; no shared bug." + "reason": "A notebook syntax error is unrelated to the _torch_extract_fbank_features runtime issue." }, { "left": "issue:46082", "right": "issue:46121", "accept": false, - "reason": "LlamaConfig head_dim validation and rope-parameter list handling are distinct config-validation issues." + "reason": "These are separate validation/config bugs: explicit head_dim rejection vs JSON-loaded list handling in rope validation." }, { "left": "issue:42994", "right": "issue:47332", "accept": false, - "reason": "Both involve saving models, but one is quantized-model saving and the other is disk-offloaded Qwen3-VL saving." + "reason": "Both involve saving, but one is quantized-model saving and the other is disk-offloaded model saving; not the same failure." }, { - "left": "issue:44016", - "right": "issue:44062", + "left": "issue:43650", + "right": "issue:44220", "accept": false, - "reason": "Notebook syntax error and AddedToken special-arg duplication are unrelated." + "reason": "The placeholder title is unrelated to the audio feature extraction bug." }, { - "left": "issue:43650", - "right": "issue:44220", + "left": "issue:44016", + "right": "issue:44062", "accept": false, - "reason": "Placeholder issue with no concrete bug vs an fbank feature-extraction failure." + "reason": "Notebook syntax error and AddedToken keyword collision are unrelated." }, { "left": "issue:43441", "right": "issue:45588", "accept": false, - "reason": "Both mention flash attention, but one is Ministral-3 model failure and the other is a sink-less-model AttributeError in the integration code." + "reason": "Both mention FlashAttention, but one is a Ministral-3 v5 RC failure and the other is an s_aux=None AttributeError in sink-less models." }, { "left": "issue:44792", "right": "issue:46424", "accept": false, - "reason": "Janus image-generation test failure and cache-length handling for quantized cache are different issues." + "reason": "A janus image-generation test failure is unrelated to cache max length being honored only for Quantized Cache." }, { "left": "issue:42907", "right": "issue:43441", "accept": false, - "reason": "Dequantized model saving and FlashAttention model failure are not the same bug." + "reason": "Both are model-specific failures, but one is save-dequantized Ministrals/Devstrals and the other is a FlashAttention issue." }, { "left": "issue:45335", "right": "issue:45584", "accept": false, - "reason": "resize_token_embeddings propagation bug vs Whisper empty-transcription generation failure." + "reason": "Different models and mechanics: token embedding resize propagation vs Whisper empty-transcription generation after align_special_tokens." }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "Video pipeline processor selection and remote_code breakage are unrelated." + "reason": "A video-classification processor lookup bug is not the same as the broad remote_code regression." }, { "left": "issue:43824", "right": "issue:44933", "accept": false, - "reason": "ImportError for a specific Qwen2.5-VL class vs a missing image_utils import; different symbols and causes." + "reason": "Both are import errors, but for different symbols/modules and likely different root causes." }, { "left": "issue:44279", "right": "issue:45405", "accept": false, - "reason": "Generic dependency issue vs a bumped MIN_PEFT_VERSION not yet released." + "reason": "Generic dependency trouble is not the same as a PEFT minimum-version bump to an unreleased release." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "Docs/behavior mismatch in text classification vs unrecognized model/config loading failure." + "reason": "Docs drift for return_all_scores is unrelated to unrecognized model/config.json issues." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "Python-version compatibility breakage and a dependency issue are not the same underlying defect." + "reason": "Python version compatibility and dependency issues are too different to treat as duplicates." }, { "left": "issue:43295", "right": "issue:44779", "accept": false, - "reason": "Processor/tokenizer regression with images vs Deepseek tokenizer output regression; different model families and failure modes." + "reason": "Different regressions: processor/tokenizer custom-code breakage vs Deepseek tokenizer output corruption." }, { - "left": "issue:43901", - "right": "issue:47217", + "left": "issue:39692", + "right": "issue:43901", "accept": false, - "reason": "TextClassificationPipeline docs issue and VL image-placeholder handling are unrelated." + "reason": "Both touch docs, but SigLIP2 example errors and TextClassificationPipeline docs drift are unrelated." }, { - "left": "issue:39692", - "right": "issue:43901", + "left": "issue:43901", + "right": "issue:47217", "accept": false, - "reason": "SigLIP2 doc example errors and return_all_scores docs mismatch are different documentation problems." + "reason": "Docs mismatch vs VL image-placeholder runtime failure are not the same bug." }, { "left": "issue:44779", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model families and likely different code paths." + "reason": "Both are tokenizer regressions, but they affect different model families and likely different tokenizer code paths." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Qwen3VL model loading regression and Qwen3.5 save_pretrained serialization regression are separate issues." + "reason": "Model loading failure and checkpoint-saving regression are different operations with different symptoms." }, { "left": "issue:42796", "right": "issue:43650", "accept": false, - "reason": "Tokenizer BPE-marker handling bug vs placeholder issue; no overlap." + "reason": "Tokenizer BPE-marker handling is unrelated to the placeholder 'ADD THE DATA' issue." }, { "left": "issue:41950", "right": "issue:45823", "accept": false, - "reason": "Video-classification processor lookup and Gemma4 PLE device mismatch are unrelated." + "reason": "Video-classification processor lookup and Gemma4 PLE device mismatch are unrelated bugs." }, { "left": "issue:43479", "right": "issue:45375", "accept": false, - "reason": "Different config classes and different bugs: default sub-config initialization vs missing field dropped by strict serialization." + "reason": "Different multimodal config bugs for different models/fields; no shared concrete code path is evident." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Serialization of tied lm_head weights vs resize_token_embeddings propagation in Gemma4." + "reason": "Serialization of tied lm_head weights and Gemma4 resize_token_embeddings propagation are separate issues." }, { "left": "issue:43577", "right": "issue:45216", "accept": false, - "reason": "Blip2 dtype propagation on load vs Qwen3.5 checkpoint serialization after save_pretrained." + "reason": "BLIP2 dtype loading behavior is unrelated to Qwen3.5 save_pretrained checkpoint corruption." }, { "left": "issue:46123", "right": "issue:46369", "accept": false, - "reason": "Partial-batch completion logic in mask generation vs memory overflow in sam3-litetext." - }, - { - "left": "issue:46614", - "right": "issue:47085", - "accept": false, - "reason": "Stale phonemizer backend selection and tokenizer thread-safety are different problems." + "reason": "A partial-batch mask pipeline bug is unrelated to sam3-litetext memory overflow." }, { "left": "issue:43653", "right": "issue:45593", "accept": false, - "reason": "BigBirdTokenizer special-token registration bug vs D-FINE auxiliary-loss behavior." + "reason": "BigBirdTokenizer special-token registration and D-FINE auxiliary-loss behavior are unrelated." } ] }, @@ -13678,14 +13627,12 @@ "issue:43441", "issue:43644", "issue:43650", - "issue:43784", "issue:43854", "issue:43873", "issue:43931", "issue:43976", "issue:44016", "issue:44075", - "issue:44186", "issue:44242", "issue:44246", "issue:44291", @@ -13714,13 +13661,16 @@ "issue:46209", "issue:46399", "issue:46512", + "issue:46614", "issue:46851", "issue:47000", + "issue:47085", "issue:47145", "issue:47217", "issue:47405" ], "soft_pairs": [ + "issue:46614|issue:47085", "issue:47145|issue:47405", "issue:43976|issue:45405", "issue:43381|issue:47000", @@ -13730,236 +13680,228 @@ "issue:43299|issue:45357", "issue:46082|issue:46399", "issue:43299|issue:43854", - "issue:44242|issue:45200", "issue:43931|issue:47145", + "issue:44242|issue:45200", + "issue:44683|issue:45542", "issue:46082|issue:46512", "issue:44016|issue:44291", - "issue:44683|issue:45542", "issue:45588|issue:47000", "issue:44351|issue:45278", "issue:43650|issue:46209", - "issue:45081|issue:45907", - "issue:45859|issue:46851", "issue:43873|issue:47145", + "issue:45859|issue:46851", + "issue:45081|issue:45907", "issue:45103|issue:45372", - "issue:45542|issue:45800", "issue:42994|issue:43317", - "issue:44336|issue:46851", + "issue:45542|issue:45800", "issue:43441|issue:45834", - "issue:44671|issue:44779", - "issue:44016|issue:45823", + "issue:44336|issue:46851", "issue:44246|issue:47217", - "issue:43784|issue:45542", - "issue:44186|issue:47000" + "issue:44671|issue:44779", + "issue:44016|issue:45823" ], - "prepared_review_unit_hash": "e19a1bd66159aa0979fc58801969d005cbdbdf6a218d9a3e7760addfe0818949", + "prepared_review_unit_hash": "81800b54c0ce08bc5f411f3ce16c00a8f725884e116ea8a12c9cbb355f2eecce", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36724, - "estimated_input_tokens": 9181, - "estimated_eval_tokens": 18618 + "soft_pair_count": 29, + "serialized_chars": 36413, + "estimated_input_tokens": 9104, + "estimated_eval_tokens": 18464 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a collection of unrelated Transformers regression reports, with a loose concentration around quantization/offload/loading issues. I did not find any soft pair that clearly describes the exact same underlying bug, so all soft edges are rejected.", - "confidence": 0.84, - "canonical_issue_reason": "issue:43873 is the best umbrella issue here: it is open, has the most discussion/inbound activity, and its quantization/offloading scope is the broadest match for the cluster\u2019s recurring model-loading problems.", + "summary": "This cluster is a loose semantic grouping of mostly unrelated Transformers v5 regressions and model-specific bugs, centered around loading, quantization/offload, tokenizers, and a few unrelated utilities. None of the soft pairs look like true duplicates.", + "confidence": 0.92, + "canonical_issue_reason": "issue:47405 is the best anchor for the cluster because it describes a high-severity, broadly framed model-loading correctness regression (silent wrong weights from an Auto-class mismatch) and is more representative than the many niche, subsystem-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the most representative and triage-worthy issue in the set because it is active, well-discussed, and centered on the dominant theme of quantization/offload behavior.", + "best_issue_reason": "issue:47405 is the strongest single issue to represent the group: it is clearly actionable, severe, and closest to the cluster\u2019s shared theme of loading/serialization regressions. The rest are mostly separate bugs rather than duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:46614", + "right": "issue:47085", + "accept": false, + "reason": "Different components and failure modes: stale phonemizer backend state vs tokenizer thread-safety regression." + }, { "left": "issue:47145", "right": "issue:47405", "accept": false, - "reason": "Both are about loading problems that can lead to bad model behavior, but one is FP8 compressed-tensors weight scaling and the other is wrong Auto-class / model_type conversion. Different code paths and failure modes." + "reason": "Both are silent-corruption-style bugs, but one is FP8 weight-scale loading and the other is wrong Auto-class conversion/loading; different code paths." }, { "left": "issue:43976", "right": "issue:45405", "accept": false, - "reason": "Python version support failure and PEFT minimum-version bump are unrelated release/dependency issues." + "reason": "One is a Python version compatibility bug; the other is an unreleased dependency version bump." }, { "left": "issue:43381", "right": "issue:47000", "accept": false, - "reason": "Gradient checkpointing in eval mode and RTDetrLoss infeasible cost matrix are unrelated functional bugs." + "reason": "Unrelated bugs: gradient checkpointing in eval mode vs RTDetr loss infeasible cost matrix." }, { "left": "issue:44462", "right": "issue:45081", "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json and Mistral regex patch crashing are different tokenizer loading bugs with no clear shared root cause." + "reason": "Tokenizer-file precedence bug vs Mistral regex patch crash; different loading paths." }, { "left": "issue:43644", "right": "issue:47145", "accept": false, - "reason": "One concerns non-persistent buffers being initialized with junk; the other concerns missing FP8 weight_scale application. Both touch loading, but not the same bug." + "reason": "Non-persistent buffer initialization bug vs FP8 compressed-tensor weight-scale loading bug." }, { "left": "issue:44075", "right": "issue:44368", "accept": false, - "reason": "SGD arguments not being used and a tie_word_embeddings warning are unrelated optimizer/config issues." + "reason": "Optimizer argument handling vs a model warning about tie_word_embeddings; not the same defect." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "Both mention Qwen models, but one is a loading regression for Qwen3VL MoE and the other is save_pretrained writing wrong visual encoder keys. Not the same defect." + "reason": "Different model families and symptoms: Qwen3VL MoE loading breakage vs Qwen3.5 save_pretrained visual-key regression." }, { "left": "issue:46082", "right": "issue:46399", "accept": false, - "reason": "LlamaConfig head_dim validation and Qwen3.5 kernel registration crash are unrelated model-specific bugs." + "reason": "Config validation for head_dim vs a kernel registration/decorator crash; separate issues." }, { "left": "issue:43299", "right": "issue:43854", "accept": false, - "reason": "Qwen3VL MoE loading and GLM-4.7-Flash test loading are different model-specific loading failures." + "reason": "Both are model-loading problems, but for different models and distinct failure mechanisms." + }, + { + "left": "issue:43931", + "right": "issue:47145", + "accept": false, + "reason": "Shape-mismatch loading failure vs FP8 checkpoint scaling corruption; different underlying bugs." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "Missing load-balancing loss when router logits are off and Gemma 4 requiring mm_token_type_ids are separate model-behavior bugs." + "reason": "MoE load-balancing loss logic vs Gemma 4 multimodal token-type defaults." }, { - "left": "issue:43931", - "right": "issue:47145", + "left": "issue:44683", + "right": "issue:45542", "accept": false, - "reason": "Shape mismatch on Qwen3-VL loading is unrelated to FP8 frozen checkpoint weight_scale handling." + "reason": "Torch/compiled attention incompatibility vs TensorFlow/tensorboard backend import issue." }, { "left": "issue:46082", "right": "issue:46512", "accept": false, - "reason": "Head-dim config validation and stale config.dtype after from_config are unrelated configuration bugs." + "reason": "Different API contracts: LlamaConfig head_dim validation vs from_config dtype/config synchronization." }, { "left": "issue:44016", "right": "issue:44291", "accept": false, - "reason": "Notebook syntax error and init_empty_weights TypeError are unrelated failures." - }, - { - "left": "issue:44683", - "right": "issue:45542", - "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 and missing TensorFlow backend handling are unrelated compatibility issues." + "reason": "Notebook syntax error vs init_empty_weights argument mismatch; no shared code path." }, { "left": "issue:45588", "right": "issue:47000", "accept": false, - "reason": "FlashAttention sink-less crash and RTDetrLoss infeasible cost matrix are different concrete code paths." + "reason": "FlashAttention sink handling crash vs RTDetrLoss infeasible cost matrix are unrelated." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "A missing HybridCache import is not the same bug as broad import errors after upgrading to 5.5.0." + "reason": "Specific missing HybridCache import vs broad import errors after upgrade; not clearly the same bug." }, { "left": "issue:43650", "right": "issue:46209", "accept": false, - "reason": "The generic 'ADD THE DATA' issue is unrelated to FP8 inference on B200." + "reason": "Placeholder issue with no bug detail vs FP8 inference failure on B200." }, { - "left": "issue:45081", - "right": "issue:45907", + "left": "issue:43873", + "right": "issue:47145", "accept": false, - "reason": "Mistral regex patch crashing and list_repo_files not forwarding revision are unrelated API bugs." + "reason": "Quantized offloading behavior vs missing weight_scale application; too different to merge." }, { "left": "issue:45859", "right": "issue:46851", "accept": false, - "reason": "CPU offload incompatibility in a rotary embedding forward method and variable-length position_ids handling are distinct Qwen3.5 issues." + "reason": "CPU-offload incompatibility in rotary embedding vs variable-length position_ids handling." }, { - "left": "issue:43873", - "right": "issue:47145", + "left": "issue:45081", + "right": "issue:45907", "accept": false, - "reason": "Both involve quantization/offloading, but one is a general offload expectation issue and the other is a specific silent FP8 checkpoint scaling bug." + "reason": "Mistral tokenizer regex patch crash vs list_repo_files revision forwarding; unrelated utilities." }, { "left": "issue:45103", "right": "issue:45372", "accept": false, - "reason": "auto_docstring annotation handling and Gemma 4 processor import failure are unrelated reflection/import problems." - }, - { - "left": "issue:45542", - "right": "issue:45800", - "accept": false, - "reason": "TensorBoard-without-TensorFlow backend detection and torch/transformers incompatibility are separate dependency issues." + "reason": "Auto-docstring annotation parsing crash vs processor loading broken by ReasoningEffort import." }, { "left": "issue:42994", "right": "issue:43317", "accept": false, - "reason": "Quantized model saving failure and dequantized model loading with device_map=auto are related broadly, but not the same underlying bug or mergeable fix." + "reason": "Quantized model saving failure vs dequantized model loading with device_map=auto; related theme, but different defect." }, { - "left": "issue:44336", - "right": "issue:46851", + "left": "issue:45542", + "right": "issue:45800", "accept": false, - "reason": "ANSI codes in loading_report and variable-length Qwen3.5 position_ids behavior are unrelated." + "reason": "TensorFlow backend detection bug vs torch/transformers version incompatibility; not duplicates." }, { "left": "issue:43441", "right": "issue:45834", "accept": false, - "reason": "FlashAttention regression in Ministral-3 and Kosmos2.5 long OCR index error are unrelated model-specific bugs." - }, - { - "left": "issue:44671", - "right": "issue:44779", - "accept": false, - "reason": "CamemBERT masked LM errors and Deepseek tokenizer regressions are separate model/tokenizer bugs." + "reason": "FlashAttention failure on Ministral-3 vs OCR long-input index error in Kosmos2.5; different models and failures." }, { - "left": "issue:44016", - "right": "issue:45823", + "left": "issue:44336", + "right": "issue:46851", "accept": false, - "reason": "Notebook syntax error and Gemma4 PLE device mismatch are unrelated." + "reason": "ANSI output formatting issue vs Qwen3.5 variable-length position_ids bug." }, { "left": "issue:44246", "right": "issue:47217", "accept": false, - "reason": "Slow imports and VL image-placeholder handling are unrelated." + "reason": "Import-time slowdown vs VL image-placeholder handling bug; unrelated." }, { - "left": "issue:43784", - "right": "issue:45542", + "left": "issue:44671", + "right": "issue:44779", "accept": false, - "reason": "A sentence-transformers import NameError and TensorFlow backend handling are different import/runtime issues." + "reason": "Different tokenizer regressions in different models: CamemBERT masked LM vs Deepseek tokenizer outputs." }, { - "left": "issue:44186", - "right": "issue:47000", + "left": "issue:44016", + "right": "issue:45823", "accept": false, - "reason": "LayoutLMv2Tokenizer NER/padding crashes and RTDetrLoss infeasible cost matrix are unrelated." + "reason": "Notebook syntax error vs Gemma4 device mismatch during forward; different layers of the stack." } ] }, @@ -13985,10 +13927,8 @@ "nodes": [ "issue:33290", "issue:42175", - "issue:42491", "issue:42617", "issue:42796", - "issue:43299", "issue:43441", "issue:43479", "issue:43525", @@ -13996,11 +13936,13 @@ "issue:43673", "issue:43723", "issue:43761", + "issue:43784", "issue:43824", "issue:43881", "issue:43901", "issue:43976", "issue:44162", + "issue:44186", "issue:44188", "issue:44246", "issue:44261", @@ -14012,11 +13954,10 @@ "issue:44589", "issue:44623", "issue:44749", - "issue:45072", - "issue:45083", "issue:45325", "issue:45362", "issue:45397", + "issue:45542", "issue:45588", "issue:45600", "issue:45656", @@ -14025,27 +13966,29 @@ "issue:46399", "issue:46655", "issue:46726", - "issue:46829", "issue:46846", "issue:46851", + "issue:47000", "issue:47145", "issue:47333", "issue:47427" ], "soft_pairs": [ + "issue:43784|issue:45542", + "issue:44186|issue:47000", "issue:42175|issue:43976", "issue:44623|issue:44749", - "issue:44623|issue:45397", "issue:45325|issue:46851", + "issue:44623|issue:45397", "issue:43673|issue:43761", - "issue:43479|issue:43525", - "issue:43525|issue:43881", "issue:43901|issue:44246", + "issue:43525|issue:43881", + "issue:43479|issue:43525", "issue:33290|issue:45656", - "issue:46399|issue:46846", "issue:43575|issue:47145", - "issue:44261|issue:44263", + "issue:46399|issue:46846", "issue:44589|issue:45362", + "issue:44261|issue:44263", "issue:44188|issue:44336", "issue:42796|issue:43441", "issue:45600|issue:45712", @@ -14058,220 +14001,211 @@ "issue:42617|issue:43976", "issue:46291|issue:46399", "issue:44297|issue:45397", - "issue:42491|issue:43299", "issue:46399|issue:46851", - "issue:46655|issue:46726", - "issue:45072|issue:46829", - "issue:45083|issue:46851" + "issue:46655|issue:46726" ], - "prepared_review_unit_hash": "be09ee70373da80a3c3e7f5c6166dbafa3755c3f375517a39901e42d5347d5be", + "prepared_review_unit_hash": "7e18e5938c559d28886a9463be5b1e739adb29f52b879f09832c0ba124b02d39", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36688, - "estimated_input_tokens": 9172, - "estimated_eval_tokens": 18600 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35756, + "estimated_input_tokens": 8939, + "estimated_eval_tokens": 18134 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous issue set, not a true duplicate cluster. The soft pairs mostly share broad themes only (tokenizers, loading, Qwen, deepspeed, etc.) but do not point to the same concrete bug or code path.", - "confidence": 0.93, - "canonical_issue_reason": "No single issue is a safe canonical for the whole set; the items span unrelated bugs, regressions, and docs/perf reports.", + "summary": "The cluster is a grab bag of unrelated Transformers issues rather than a tight duplicate set. The soft-similarity pairs are mostly false positives: they share broad subsystems or themes, but not the same concrete bug or change path. No PRs were provided.", + "confidence": 0.87, + "canonical_issue_reason": "issue:43824 is the strongest standalone anchor: it has a concrete import failure, multiple inbound references, and substantial discussion, making it the most actionable representative issue in an otherwise heterogeneous set.", "canonical_pr_reason": null, - "best_issue_reason": "None of the issues cleanly represents the others as a duplicate anchor; at best they are loosely related by subsystem or model family.", + "best_issue_reason": "issue:43824 is the best global issue candidate because it describes a specific, user-facing regression with clear reproduction and ecosystem impact, unlike many of the other items that are narrower or orthogonal.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Both are import/backend failures, but they are different bugs: sentence-transformers import hits a missing nn symbol, while tensorboard-only installs trigger a missing TensorFlow backend path." + }, + { + "left": "issue:44186", + "right": "issue:47000", + "accept": false, + "reason": "Completely different failures: a tokenizer crash on NER/padding versus an RTDetrLoss infeasible cost-matrix error." + }, { "left": "issue:42175", "right": "issue:43976", "accept": false, - "reason": "Different problems: backend extras packaging vs Python version compatibility." + "reason": "One is a packaging/backend omission for pip extras, the other is a Python version compatibility problem; not the same underlying issue." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Missing processor save files is unrelated to data preprocessing slowdown after upgrade." + "reason": "Saving processor files and post-upgrade performance regression are unrelated." }, { - "left": "issue:44623", - "right": "issue:45397", + "left": "issue:45325", + "right": "issue:46851", "accept": false, - "reason": "Different save/load paths and different model families; not the same bug." + "reason": "Both mention position_ids, but they affect different Qwen models and different code paths: still-image temporal scaling vs variable-length input handling." }, { - "left": "issue:45325", - "right": "issue:46851", + "left": "issue:44623", + "right": "issue:45397", "accept": false, - "reason": "Both involve position_ids, but they affect different Qwen models and distinct logic paths." + "reason": "Processor save-pretrained file omission is unrelated to a Zero3 from_pretrained loading failure." }, { "left": "issue:43673", "right": "issue:43761", "accept": false, - "reason": "Cache handling in generation is unrelated to CLIP hidden_states regression." + "reason": "Generation cache handling during chunked prefill is unrelated to CLIPVisionModel hidden_states being omitted." }, { - "left": "issue:43479", - "right": "issue:43525", + "left": "issue:43901", + "right": "issue:44246", "accept": false, - "reason": "Config default initialization bug vs missing pad_token_id attribute; different failures." + "reason": "Docs mismatch for return_all_scores is not the same problem as intermittent slow imports." }, { "left": "issue:43525", "right": "issue:43881", "accept": false, - "reason": "Different model/config errors with no shared concrete code-path." + "reason": "Missing pad_token_id on Llama4Config and a glm-4v-9b load failure are different model-specific bugs." }, { - "left": "issue:43901", - "right": "issue:44246", + "left": "issue:43479", + "right": "issue:43525", "accept": false, - "reason": "Docs mismatch for a pipeline option is unrelated to intermittent slow imports." + "reason": "Default multimodal config initialization and missing pad_token_id are separate config issues." }, { "left": "issue:33290", "right": "issue:45656", "accept": false, - "reason": "Both mention deepspeed, but one is OOM with Adafactor and the other is double optimizer stepping." - }, - { - "left": "issue:46399", - "right": "issue:46846", - "accept": false, - "reason": "Kernel registration crash is unrelated to missing TP plan entries and shape/OOM issues." + "reason": "Both involve deepspeed/optimizer behavior, but one is OOM with Adafactor and the other is optimizer.step() being invoked twice; not mergeable as one fix." }, { "left": "issue:43575", "right": "issue:47145", "accept": false, - "reason": "TP OOM on model load is unrelated to FP8 weight_scale being skipped on frozen checkpoints." + "reason": "Tensor-parallel OOM on Qwen2-57B is unrelated to silent FP8 checkpoint corruption and missing weight_scale application." }, { - "left": "issue:44261", - "right": "issue:44263", + "left": "issue:46399", + "right": "issue:46846", "accept": false, - "reason": "Different model internals: RMS norm epsilon precision issue vs torch.split indexing semantics." + "reason": "Different Qwen3.5 failures: kernel registration/decorator error versus missing TP-plan entries causing OOM/shape errors." }, { "left": "issue:44589", "right": "issue:45362", "accept": false, - "reason": "Storage type lookup failure is unrelated to chat-time Qwen3.5 crashing." + "reason": "A storage-object lookup TypeError is unrelated to a chat crash on Qwen3.5." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Different model internals and failures: MLA normalization precision/config issue versus torch.split return handling." }, { "left": "issue:44188", "right": "issue:44336", "accept": false, - "reason": "Attention kernel divergence is unrelated to ANSI codes in loading_report." + "reason": "Attention-kernel divergence under torch.compile is unrelated to ANSI codes in loading_report." }, { "left": "issue:42796", "right": "issue:43441", "accept": false, - "reason": "Same model family, but tokenizer BPE handling and FlashAttention failure are different bugs." + "reason": "Tokenizer BPE-marker handling and FlashAttention support are different bugs, even though both involve Ministral-3." }, { "left": "issue:45600", "right": "issue:45712", "accept": false, - "reason": "Broken auto_mappings references and leftover dummy classes are different CI hygiene issues." + "reason": "Removed auto-mapping references breaking CI and leftover dummy PT classes are separate cleanup problems." }, { "left": "issue:45588", "right": "issue:47427", "accept": false, - "reason": "FlashAttention s_aux handling is unrelated to redundant init on non-CUDA hardware." + "reason": "Flash-attention sink handling and redundant HF init on non-CUDA hardware are different code paths." }, { "left": "issue:45362", "right": "issue:47333", "accept": false, - "reason": "Chat crash vs disk-offload save failure are separate code paths." + "reason": "Qwen3.5 chat crash is unrelated to saving a disk-offloaded Qwen3-30B-A3B model." }, { "left": "issue:43824", "right": "issue:43976", "accept": false, - "reason": "Missing import for a VL class is unrelated to Python 3.9/3.10 compatibility." + "reason": "An import error for a specific model class is unrelated to a Python version support issue." }, { "left": "issue:44246", "right": "issue:44496", "accept": false, - "reason": "Slow imports are unrelated to unrecognized model config loading." + "reason": "Intermittent import slowness and unrecognized-model config parsing are not the same bug." }, { "left": "issue:44336", "right": "issue:44373", "accept": false, - "reason": "Loading-report ANSI formatting and a docstring typo are not the same bug." + "reason": "ANSI output leakage and a wrong docstring are unrelated." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "Tokenizer loading in v5 is unrelated to ESM2 breakage." + "reason": "Tokenizer loading in v5 and ESM2 breakage are different failures with no shared concrete code path." }, { "left": "issue:42617", "right": "issue:43976", "accept": false, - "reason": "3D parallel example failure is unrelated to Python-version support." + "reason": "Running 3d_parallel.py failing and Python 3.9 compatibility are unrelated." }, { "left": "issue:46291", "right": "issue:46399", "accept": false, - "reason": "Kernel dependency/version import error is not the same as the rotary-pos-emb registration crash." + "reason": "kernels package metadata requirements and a missing decorator in Qwen3.5 kernel registration are different issues." }, { "left": "issue:44297", "right": "issue:45397", "accept": false, - "reason": "Tokenizer save metadata bug is unrelated to gemma-4 zero3 loading." - }, - { - "left": "issue:42491", - "right": "issue:43299", - "accept": false, - "reason": "Both involve Qwen MoE regressions, but the model variants and failure modes differ too much to merge." + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to gemma-4 Zero3 loading." }, { "left": "issue:46399", "right": "issue:46851", "accept": false, - "reason": "Kernel registration failure is unrelated to variable-length position_ids handling." + "reason": "Kernel registration failure and variable-length position_ids handling are separate Qwen3.5 problems." }, { "left": "issue:46655", "right": "issue:46726", "accept": false, - "reason": "Video padding calculation bug is unrelated to backend-missing image processor errors." - }, - { - "left": "issue:45072", - "right": "issue:46829", - "accept": false, - "reason": "BF16 dtype mismatch bugs are unrelated to hidden_states packaging in classification output." - }, - { - "left": "issue:45083", - "right": "issue:46851", - "accept": false, - "reason": "Different Qwen model subsystems and different symptoms; not the same underlying bug." + "reason": "Video padding calculation and missing vision-backend image processor resolution are different processor bugs." } ] }, @@ -14297,29 +14231,28 @@ "nodes": [ "issue:36246", "issue:41950", + "issue:42491", "issue:42915", "issue:43257", - "issue:43295", - "issue:43352", + "issue:43299", "issue:43526", "issue:43540", "issue:43550", "issue:43596", - "issue:43901", "issue:44220", "issue:44242", "issue:44246", "issue:44295", "issue:44336", "issue:44479", + "issue:44496", "issue:44779", "issue:44871", - "issue:44995", "issue:45003", + "issue:45072", + "issue:45083", "issue:45103", "issue:45137", - "issue:45200", - "issue:45250", "issue:45292", "issue:45375", "issue:45397", @@ -14338,6 +14271,7 @@ "issue:46726", "issue:46752", "issue:46821", + "issue:46829", "issue:46851", "issue:47176", "issue:47177", @@ -14345,6 +14279,9 @@ "issue:47312" ], "soft_pairs": [ + "issue:45072|issue:46829", + "issue:42491|issue:43299", + "issue:45083|issue:46851", "issue:47176|issue:47177", "issue:45003|issue:47217", "issue:43526|issue:45003", @@ -14354,222 +14291,219 @@ "issue:45103|issue:46399", "issue:44295|issue:45397", "issue:45561|issue:45563", - "issue:43550|issue:46506", "issue:44336|issue:46424", + "issue:43550|issue:46506", "issue:43540|issue:46752", - "issue:46614|issue:46665", "issue:36246|issue:43596", + "issue:46614|issue:46665", "issue:42915|issue:47312", - "issue:46726|issue:47217", "issue:43540|issue:46655", + "issue:46726|issue:47217", "issue:43257|issue:45137", "issue:44779|issue:45292", "issue:44479|issue:46655", - "issue:43901|issue:45003", + "issue:44496|issue:45003", "issue:41950|issue:44220", "issue:45375|issue:45759", "issue:45292|issue:45701", - "issue:44242|issue:45593", - "issue:43295|issue:45292", - "issue:43352|issue:45250", - "issue:44995|issue:45200" + "issue:44242|issue:45593" ], - "prepared_review_unit_hash": "b03e4b87486e30d6a9565159069091b988bf73b2d054fe8de4e386f7de0c83da", + "prepared_review_unit_hash": "e639f8d3dc4ea1331d995e396bc0b022f965ce8853b43758027b45942a92b19e", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36019, - "estimated_input_tokens": 9005, - "estimated_eval_tokens": 18266 + "serialized_chars": 36177, + "estimated_input_tokens": 9045, + "estimated_eval_tokens": 18346 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab bag of largely unrelated issue reports across security, pipelines, multimodal/Qwen, DeepSpeed, docs, and model-specific bugs. None of the soft edges look like the same concrete underlying bug or change, so no duplicates are accepted.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: a few pairs share model family or subsystem vocabulary, but none look like the same underlying bug/change closely enough to merge as duplicates. No PRs are present.", + "confidence": 0.9, + "canonical_issue_reason": "No single issue is a safe canonical duplicate target; the set mixes unrelated bugs across Qwen/VL, distributed loading, tokenizer/config, import-time, security, and utility code paths.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:45003 is the closest broad/core-library candidate and appears in multiple soft-similarity links, but it still does not represent the rest of the cluster as a duplicate.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45072", + "right": "issue:46829", + "accept": false, + "reason": "Different failures: one is a bfloat16 dtype mismatch in model inference, the other is an output field/hidden_states mix-up in VideoPrism classification." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen3 MoE loading, but one is LoRA compatibility across versions and the other is accelerate/deepspeed weight conversion; not the same bug." + }, + { + "left": "issue:45083", + "right": "issue:46851", + "accept": false, + "reason": "Both involve Qwen3.5 positional/length behavior, but one is a helper output-length issue and the other is variable-length position_ids handling." + }, { "left": "issue:47176", "right": "issue:47177", "accept": false, - "reason": "Different security bugs: command-substitution allowlist bypass vs tar/zip-slip arbitrary file write." + "reason": "Distinct security issues: shell command-substitution bypass vs tar/zip-slip arbitrary file write." }, { "left": "issue:45003", "right": "issue:47217", "accept": false, - "reason": "Unrelated failures: unsafe sys.modules access vs vision-language placeholder handling in user text." + "reason": "sys.modules access in modeling_utils is unrelated to VL image placeholder handling in user text." }, { "left": "issue:43526", "right": "issue:45003", "accept": false, - "reason": "BEiT label reduction bug is unrelated to the sys.modules access issue." + "reason": "BEiT image-processor label reduction and unsafe sys.modules access are unrelated code paths." }, { "left": "issue:45588", "right": "issue:46851", "accept": false, - "reason": "Different code paths: flash-attention sinkless-model crash vs variable-length position_ids handling." + "reason": "Flash-attention s_aux=None crash is a multimodal attention bug, not the same as Qwen3.5 variable-length position_ids handling." }, { "left": "issue:45003", "right": "issue:46821", "accept": false, - "reason": "Unrelated: sys.modules lookup bug vs TimesFM window_size attribute error." + "reason": "Unsafe sys.modules access and TimesFM window_size AttributeError are different core-library bugs." }, { "left": "issue:44246", "right": "issue:44871", "accept": false, - "reason": "Different model/config problems: missing router aux loss vs eos_token_id mismatch." + "reason": "Load-balancing loss omission and eos_token_id config inconsistency affect different model behaviors and code paths." }, { "left": "issue:45103", "right": "issue:46399", "accept": false, - "reason": "auto_docstring annotations crash is unrelated to Qwen rotary registration/decorator failure." + "reason": "Auto-docstring future-annotations crash is unrelated to the Qwen3.5 kernel registration error." }, { "left": "issue:44295", "right": "issue:45397", "accept": false, - "reason": "Different failures: position_ids buffer access vs Gemma-4 Zero3 from_pretrained issue." + "reason": "Reading position_ids after buffer registration and gemma-4 ZeRO3 from_pretrained failures are different distributed/model-loading problems." }, { "left": "issue:45561", "right": "issue:45563", "accept": false, - "reason": "pytest-xdist captured_info race and stale num_return_sequences warning are separate bugs." - }, - { - "left": "issue:43550", - "right": "issue:46506", - "accept": false, - "reason": "Both mention runtime behavior, but one is torch.compile/SDPA and the other CPU vs CUDA output divergence." + "reason": "pytest-xdist file race and stale num_return_sequences warning are unrelated testing/generate issues." }, { "left": "issue:44336", "right": "issue:46424", "accept": false, - "reason": "ANSI color output in loading_report is unrelated to cache max_cache_length behavior." + "reason": "ANSI leakage in loading_report and cache max_length behavior are unrelated utility vs cache-config bugs." }, { - "left": "issue:43540", - "right": "issue:46752", + "left": "issue:43550", + "right": "issue:46506", "accept": false, - "reason": "Video-input ValueError in Qwen3OmniMoe and empty-list chat template IndexError are unrelated." + "reason": "torch.compile SDPA failure in Bamba and CPU/CUDA divergence in PPDocLayoutV3 are different backend/model issues." }, { - "left": "issue:46614", - "right": "issue:46665", + "left": "issue:43540", + "right": "issue:46752", "accept": false, - "reason": "Stale phonemizer backend language is unrelated to from_pretrained behavior inside classes." + "reason": "Qwen3OmniMoe video-input ValueError and empty-list chat template crash are unrelated failures." }, { "left": "issue:36246", "right": "issue:43596", "accept": false, - "reason": "Import error for Qwen2_5_VLImageProcessor and DeepSpeed Zero3 BertModel index error are unrelated." + "reason": "ImportError for Qwen2_5_VLImageProcessor and DeepSpeed ZeRO3 BertModel IndexError are different bugs." }, { - "left": "issue:42915", - "right": "issue:47312", + "left": "issue:46614", + "right": "issue:46665", "accept": false, - "reason": "Qwen3Moe FP8 config failure is unrelated to kernels 0.16 compatibility." + "reason": "Stale phonemizer backend state and from_pretrained behavior inside Python classes do not describe the same issue." }, { - "left": "issue:46726", - "right": "issue:47217", + "left": "issue:42915", + "right": "issue:47312", "accept": false, - "reason": "Missing vision backend/unrecognized image processor is different from image-placeholder-in-user-text failure." + "reason": "Qwen3Moe FineGrainedFP8Config failure and kernels 0.16 incompatibility are different compatibility problems." }, { "left": "issue:43540", "right": "issue:46655", "accept": false, - "reason": "Both are video-processing issues, but they affect different models and appear to be different bugs." + "reason": "Both are video-related, but Qwen3OmniMoe ValueError and SmolVLM padding miscalculation are distinct model-specific bugs." + }, + { + "left": "issue:46726", + "right": "issue:47217", + "accept": false, + "reason": "Missing vision backend / unrecognized image processor is not the same as failing when the image placeholder appears in user text." }, { "left": "issue:43257", "right": "issue:45137", "accept": false, - "reason": "Qwen3 weight conversion with accelerate/deepspeed is unrelated to the ZeRO3 empty-deque error." + "reason": "MoE weight conversion with accelerate+deepspeed and ZeRO3 empty-deque pop are different distributed-loading failures." }, { "left": "issue:44779", "right": "issue:45292", "accept": false, - "reason": "Tokenizer regression and resize_token_embeddings/output_embeddings mismatch are different problems." + "reason": "Tokenizer changes across versions and resize_token_embeddings not updating output_embeddings are unrelated." }, { "left": "issue:44479", "right": "issue:46655", "accept": false, - "reason": "Both involve video inputs, but the broad Qwen VL regression and SmolVLM padding bug are not the same concrete issue." + "reason": "General v5.3.0 video-input regression across several VL models and SmolVLM padding calculation are related only at a very broad level, not the same concrete bug." }, { - "left": "issue:43901", + "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "Docs mismatch for return_all_scores is unrelated to sys.modules access." + "reason": "Unrecognized model/model_type config error and unsafe sys.modules access are unrelated loading-time issues." }, { "left": "issue:41950", "right": "issue:44220", "accept": false, - "reason": "Video-classification image-processor lookup and _torch_extract_fbank_features are separate pipeline bugs." + "reason": "Video-classification pipeline choosing image processors and _torch_extract_fbank_features() are different pipeline defects." }, { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "Missing deepstack_visual_indexes and not unwrapping text_config are different composite-Qwen config issues." + "reason": "Missing deepstack_visual_indexes field and composite Qwen multimodal text_config unwrapping are distinct config-handling bugs." }, { "left": "issue:45292", "right": "issue:45701", "accept": false, - "reason": "resize_token_embeddings/output_embeddings bug is unrelated to tokenization changes across versions." + "reason": "resize_token_embeddings not affecting output_embeddings and version-dependent tokenization changes are different behaviors." }, { "left": "issue:44242", "right": "issue:45593", "accept": false, - "reason": "Missing load-balancing loss and missing auxiliary losses in D-FINE are related only by general 'loss not added' wording, not the same bug." - }, - { - "left": "issue:43295", - "right": "issue:45292", - "accept": false, - "reason": "Processor.tokenizer regression is unrelated to resize_token_embeddings not updating output embeddings." - }, - { - "left": "issue:43352", - "right": "issue:45250", - "accept": false, - "reason": "Model-specific FA2 unsupported error is not the same as the generic Flash Attention 2.0 topic." - }, - { - "left": "issue:44995", - "right": "issue:45200", - "accept": false, - "reason": "GlmMoeDsa stale cache crash and Gemma 4 mm_token_type_ids defaulting issue are unrelated." + "reason": "Load-balancing loss gated on output_router_logits and D-FINE auxiliary-loss omission with denoising off are separate model-loss bugs." } ] }, @@ -14598,30 +14532,29 @@ "issue:36246", "issue:42757", "issue:42886", + "issue:43295", + "issue:43352", "issue:43475", "issue:43540", - "issue:43653", "issue:44038", "issue:44060", "issue:44077", "issue:44162", - "issue:44242", "issue:44295", "issue:44351", - "issue:44368", "issue:44451", "issue:44466", "issue:44479", "issue:44623", "issue:44749", - "issue:44857", "issue:44877", - "issue:44977", "issue:44987", "issue:44995", "issue:45003", "issue:45137", "issue:45200", + "issue:45250", + "issue:45292", "issue:45335", "issue:45375", "issue:45397", @@ -14643,224 +14576,224 @@ "issue:47429" ], "soft_pairs": [ + "issue:43295|issue:45292", + "issue:43352|issue:45250", + "issue:44995|issue:45200", "issue:36246|issue:44351", "issue:36246|issue:42757", - "issue:44466|issue:45335", "issue:44060|issue:44077", - "issue:46170|issue:47321", "issue:45137|issue:45600", + "issue:46170|issue:47321", + "issue:44466|issue:45335", "issue:44162|issue:45200", "issue:42886|issue:46132", "issue:44295|issue:44749", - "issue:44451|issue:46752", "issue:44995|issue:45588", - "issue:44877|issue:45375", + "issue:44451|issue:46752", "issue:45685|issue:46431", + "issue:44877|issue:45375", "issue:44479|issue:46170", "issue:47217|issue:47429", "issue:43475|issue:46369", "issue:45003|issue:45561", "issue:44877|issue:46170", - "issue:44987|issue:45459", "issue:44623|issue:46726", + "issue:44987|issue:45459", "issue:44038|issue:46506", - "issue:45397|issue:45656", "issue:29870|issue:30333", - "issue:44242|issue:44857", - "issue:43475|issue:43540", - "issue:43653|issue:44162", - "issue:44368|issue:44977" + "issue:45397|issue:45656", + "issue:43475|issue:43540" ], - "prepared_review_unit_hash": "1f126e030b39724ca7317841928c0a0012478fa7661d02d0c7e462971825a589", + "prepared_review_unit_hash": "a57951c26c2be290aefedabcd3aebc240fc15a01ecef0310828faac39a2ca9ba", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 27, - "serialized_chars": 35820, - "estimated_input_tokens": 8955, - "estimated_eval_tokens": 18166 + "serialized_chars": 35126, + "estimated_input_tokens": 8782, + "estimated_eval_tokens": 17820 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: the soft pairs share only broad vocabulary (imports, configs, vision/video, deepspeed, tied weights) but point to different code paths and different fixes. I would not merge any of the proposed soft pairs as duplicates.", - "confidence": 0.89, - "canonical_issue_reason": "If one issue must represent the cluster, issue 44479 is the broadest user-facing regression (video input breakage across multiple Qwen variants), but overall there is no true single canonical duplicate here.", + "summary": "This set is not a true duplicate cluster: it contains many unrelated bug reports, with only superficial overlap in words like \"tokenizer\", \"video\", \"config\", or \"deepspeed\". None of the soft pairs look like the same underlying bug/change, so no dedupes should be merged.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue is appropriate here; the items are heterogeneous and do not collapse to one underlying bug.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44479 is the strongest representative because it spans several related model variants and describes a concrete regression, unlike the mostly isolated model-specific import/config bugs around it.", + "best_issue_reason": "No global best issue: selecting one would be arbitrary because the cluster is not coherent.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "Both mention embedding/processor regressions, but one is about custom multimodal processor.tokenizer behavior and the other about resize_token_embeddings not updating output embeddings; different code paths and failure modes." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "43352 is a specific model gate for NemotronHForCausalLM, while 45250 is just a general Flash Attention topic; not the same bug." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "Stale indexer cache crash on second forward is unrelated to Gemma 4 requiring mm_token_type_ids to default to zeros." + }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different symbols and likely unrelated failure points: Qwen2.5 VLImageProcessor vs HybridCache." + "reason": "Both are import errors, but for different missing symbols from different modules and versions; not the same underlying issue." }, { "left": "issue:36246", "right": "issue:42757", "accept": false, - "reason": "Different missing imports from different packages; one is transformers model export, the other is huggingface_hub offline helper." + "reason": "Qwen2_5_VLImageProcessor import failure is unrelated to huggingface_hub.is_offline_mode missing." }, { - "left": "issue:44466", - "right": "issue:45335", + "left": "issue:44060", + "right": "issue:44077", "accept": false, - "reason": "Both involve weight/tied-embedding behavior, but one is device-dependent serialization of lm_head.weight and the other is resize_token_embeddings not updating decoder.embed_tokens." + "reason": "One is a tied-weights warning bug in Qwen3-Next; the other is a model policy change for patchtsmixer post_init. Different areas and fixes." }, { - "left": "issue:44060", - "right": "issue:44077", + "left": "issue:45137", + "right": "issue:45600", "accept": false, - "reason": "Different problems: a tied-weights warning in Qwen3-Next versus PatchTSMixer post_init API/behavior." + "reason": "DeepSpeed ZeRO3 deque underflow and auto_mappings referencing removed Sam3LiteText configs are unrelated." }, { "left": "issue:46170", "right": "issue:47321", "accept": false, - "reason": "Both mention AMD-related setups, but one is a Windows ROCm/torch.distributed import error and the other is a quark class update issue." + "reason": "Windows ROCm/torch.distributed import problems are unrelated to AMD quark class updates." }, { - "left": "issue:45137", - "right": "issue:45600", + "left": "issue:44466", + "right": "issue:45335", "accept": false, - "reason": "DeepSpeed ZeRO3 deque underflow is unrelated to auto_mappings referencing removed Sam3LiteText configs." + "reason": "Both involve resizing/tied embeddings, but one is serialization of lm_head.weight and the other is decoder.embed_tokens not updating; distinct bugs." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "Different model families and symptoms: ESM2 breakage versus Gemma 4 mm_token_type_ids defaulting for text-only fine-tuning." + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids behavior are completely different models and code paths." }, { "left": "issue:42886", "right": "issue:46132", "accept": false, - "reason": "Offline tokenizer cache loading and AttentionInterface.register behavior are unrelated subsystems." + "reason": "Tokenizer cache loading in offline mode is unrelated to AttentionInterface.register changing function behavior." }, { "left": "issue:44295", "right": "issue:44749", "accept": false, - "reason": "Registering position_ids as a buffer is not the same bug as a generic performance slowdown after upgrading Transformers." - }, - { - "left": "issue:44451", - "right": "issue:46752", - "accept": false, - "reason": "Loading a specific model failing is unrelated to apply_chat_template crashing on an empty list." + "reason": "Reading position_ids after buffer registration and a large slowdown after upgrading transformers are different regressions." }, { "left": "issue:44995", "right": "issue:45588", "accept": false, - "reason": "A stale indexer cache crash in GlmMoeDsa is unrelated to an s_aux=None flash-attention AttributeError." + "reason": "Stale indexer cache on second forward is unrelated to flash_attention s_aux=None AttributeError." }, { - "left": "issue:44877", - "right": "issue:45375", + "left": "issue:44451", + "right": "issue:46752", "accept": false, - "reason": "Strict config loading for granite_speech and missing deepstack_visual_indexes in Qwen3.5MoeVisionConfig are different config issues." + "reason": "Failed model load for ScandiBERT and apply_chat_template crashing on empty input are unrelated." }, { "left": "issue:45685", "right": "issue:46431", "accept": false, - "reason": "MPS histogram support for Int tensors is unrelated to missing chat templates in processors." + "reason": "MPS histogram Int error in moe code is unrelated to missing chat template on a processor." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both mention strict config handling, but one blocks loading granite_speech and the other drops a missing field in Qwen3_5MoeVisionConfig; not the same artifact or fix." }, { "left": "issue:44479", "right": "issue:46170", "accept": false, - "reason": "Video-input regression in Qwen models is unrelated to Windows ROCm/torch.distributed import failures." + "reason": "Qwen video-input regression and Windows ROCm import/type errors are unrelated." }, { "left": "issue:47217", "right": "issue:47429", "accept": false, - "reason": "One is a VL placeholder-handling bug; the other is a tokenizers version upper-bound packaging issue." + "reason": "VL image-placeholder parsing and a tokenizers version upper-bound issue are unrelated." }, { "left": "issue:43475", "right": "issue:46369", "accept": false, - "reason": "Missing fpn_position_embeddings in SAM 3 Video is unrelated to memory overflow in sam3-litetext." + "reason": "Both are SAM3-family, but one is a missing attribute in video vision encoder output and the other is a memory overflow in sam3-litetext; not the same bug." }, { "left": "issue:45003", "right": "issue:45561", "accept": false, - "reason": "Unsafe sys.modules access is a different bug from the xdist race on captured_info.txt." + "reason": "Unsafe sys.modules access and pytest-xdist race on captured_info.txt are different concurrency-related problems." }, { "left": "issue:44877", "right": "issue:46170", "accept": false, - "reason": "These are unrelated: config strictness for granite_speech versus Windows ROCm import/type errors." - }, - { - "left": "issue:44987", - "right": "issue:45459", - "accept": false, - "reason": "Model loading failure for physical-intelligence/fast is not the same as tokenizer error masking when protobuf is absent." + "reason": "granite_speech strict config loading and Windows ROCm torch.distributed import issues are unrelated." }, { "left": "issue:44623", "right": "issue:46726", "accept": false, - "reason": "processor.save_pretrained missing files is different from AutoProcessor reporting an unrecognized image processor due to missing backend." + "reason": "Missing files from processor.save_pretrained and unrecognized image processor when a vision backend is missing are different failure modes." }, { - "left": "issue:44038", - "right": "issue:46506", + "left": "issue:44987", + "right": "issue:45459", "accept": false, - "reason": "A Qwen3-VL-Moe bug and CPU/CUDA result divergence in PPDocLayoutV3 are different concrete code paths." + "reason": "Model loading failure for physical-intelligence/fast and protobuf-related tokenizer error masking are unrelated." }, { - "left": "issue:45397", - "right": "issue:45656", + "left": "issue:44038", + "right": "issue:46506", "accept": false, - "reason": "gemma-4 zero3 from_pretrained and a duplicate optimizer step under deepspeed are separate training/runtime bugs." + "reason": "Qwen3-VL-Moe bug and CPU/CUDA result divergence in PPDocLayoutV3 are different models and code paths." }, { "left": "issue:29870", "right": "issue:30333", "accept": false, - "reason": "Bart generate() UnboundLocalError and MLflow job-state reporting are unrelated." + "reason": "Bart generate UnboundLocalError and MLflow job status reporting are unrelated." }, { - "left": "issue:44242", - "right": "issue:44857", + "left": "issue:45397", + "right": "issue:45656", "accept": false, - "reason": "Missing load-balancing loss when output_router_logits=False is unrelated to an AMP/CUDA crash in LwDetrImageLoss." + "reason": "Gemma-4 ZeRO3 from_pretrained issues and a double optimizer step with deepspeed are separate bugs." }, { "left": "issue:43475", "right": "issue:43540", "accept": false, - "reason": "SAM 3 Video missing an attribute and Qwen3OmniMoe video-input validation failures are different model pipelines." - }, - { - "left": "issue:43653", - "right": "issue:44162", - "accept": false, - "reason": "BigBirdTokenizer special-token decode behavior is unrelated to ESM2 being broken." - }, - { - "left": "issue:44368", - "right": "issue:44977", - "accept": false, - "reason": "A tied-weights warning in Qwen3.5 fine-tuning is not the same as a flash-attention generation regression." + "reason": "Both are multimodal/video-related, but one is a missing attribute in SAM3 output and the other is a Qwen3OmniMoe video-processing ValueError; different models and fixes." } ] }, @@ -14884,13 +14817,12 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ - "issue:39692", "issue:42199", "issue:42886", "issue:42994", "issue:43065", "issue:43408", - "issue:43475", + "issue:43653", "issue:43701", "issue:43723", "issue:43784", @@ -14898,13 +14830,15 @@ "issue:43931", "issue:43976", "issue:43994", + "issue:44162", "issue:44188", - "issue:44297", + "issue:44242", "issue:44361", "issue:44368", "issue:44483", - "issue:44623", "issue:44749", + "issue:44857", + "issue:44977", "issue:45103", "issue:45229", "issue:45412", @@ -14923,7 +14857,6 @@ "issue:46710", "issue:46726", "issue:46752", - "issue:46821", "issue:46851", "issue:46858", "issue:47020", @@ -14934,238 +14867,231 @@ "issue:47429" ], "soft_pairs": [ + "issue:44242|issue:44857", + "issue:44368|issue:44977", + "issue:43653|issue:44162", "issue:46305|issue:46315", "issue:42994|issue:47312", - "issue:43723|issue:46710", "issue:43065|issue:43408", + "issue:43723|issue:46710", "issue:45412|issue:46123", "issue:42886|issue:45229", "issue:47059|issue:47429", "issue:43931|issue:44368", "issue:43906|issue:46202", "issue:45987|issue:46082", - "issue:44361|issue:44749", "issue:45103|issue:46858", + "issue:44361|issue:44749", "issue:42199|issue:46851", "issue:43994|issue:46726", "issue:44368|issue:45563", - "issue:44188|issue:46202", "issue:46132|issue:47145", "issue:46752|issue:47020", - "issue:43701|issue:44749", + "issue:44188|issue:46202", "issue:45593|issue:47068", "issue:43976|issue:47429", + "issue:43701|issue:44749", "issue:44483|issue:46202", "issue:43784|issue:46399", "issue:44188|issue:46421", - "issue:44749|issue:45656", - "issue:44297|issue:44623", - "issue:39692|issue:46726", - "issue:43065|issue:46123", - "issue:43475|issue:46821" + "issue:44749|issue:45656" ], - "prepared_review_unit_hash": "5a97b4ed62100e788416cabc50ae724f02f590ff37f21d78d2e750f1f711557a", + "prepared_review_unit_hash": "7fa01b1d0b4596fde07b06a4579234fc3a78ebfeb72346797d21a4073e1f134c", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36641, - "estimated_input_tokens": 9161, - "estimated_eval_tokens": 18578 + "soft_pair_count": 28, + "serialized_chars": 36278, + "estimated_input_tokens": 9070, + "estimated_eval_tokens": 18396 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous, and every soft edge looks like a false positive rather than a true duplicate. The items span unrelated tokenizer, model-loading, cache, compile, pipeline, and documentation bugs.", + "summary": "These issues are largely unrelated; none of the soft pairs look like the same underlying bug or change, so I rejected all candidate duplicates. The strongest standalone issue in the set is 45229 because it has the richest discussion and the clearest bug context.", "confidence": 0.96, - "canonical_issue_reason": "No canonical issue: these items cover unrelated bugs across tokenizers, cache, models, pipelines, and docs.", + "canonical_issue_reason": "Issue 45229 is the most complete and high-signal bug report in the set, with extensive discussion and a concrete multi-GPU CUDA OOM reproduction.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a safe global representative; the soft pairs are all non-duplicates.", + "best_issue_reason": "Issue 45229 is the best overall issue candidate because it is specific, well-documented, and has the most diagnostic context.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Different problems: MoE load-balancing loss vs D-FINE image loss/AMP crash." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated Qwen3.5 warning about tie_word_embeddings vs flash-attention generation failure." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token bug is unrelated to ESM2 model breakage." + }, { "left": "issue:46305", "right": "issue:46315", "accept": false, - "reason": "Different failure modes: a streaming mask off-by-one versus a tokenizer conversion slowdown regression." + "reason": "Sliding-window mask off-by-one is unrelated to tokenizer conversion slowdown." }, { "left": "issue:42994", "right": "issue:47312", "accept": false, - "reason": "Quantized model saving failure is unrelated to kernel compatibility with version 0.16." + "reason": "Quantized model saving failure is unrelated to kernel version incompatibility." }, { - "left": "issue:43723", - "right": "issue:46710", + "left": "issue:43065", + "right": "issue:43408", "accept": false, - "reason": "Tokenizer loading in v5 and an accuracy regression after upgrade are different problems." + "reason": "Both touch SAM3, but one is a dummy Conv2d concern and the other is a model-type warning; not the same bug." }, { - "left": "issue:43065", - "right": "issue:43408", + "left": "issue:43723", + "right": "issue:46710", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as a model-type mismatch warning." + "reason": "Tokenizer loading in v5 is unrelated to an accuracy regression in DeepSeek-R1." }, { "left": "issue:45412", "right": "issue:46123", "accept": false, - "reason": "RT-DETR memory cleanup and MaskGenerationPipeline partial-batch handling are unrelated." + "reason": "Memory not released in RT-DETR is unrelated to partial-batch result dropping in MaskGenerationPipeline." }, { "left": "issue:42886", "right": "issue:45229", "accept": false, - "reason": "Offline tokenizer cache loading is unrelated to Gemma multi-GPU CUDA OOM." + "reason": "Offline tokenizer cache loading failure is unrelated to Gemma4 multi-GPU OOM." }, { "left": "issue:47059", "right": "issue:47429", "accept": false, - "reason": "A CLI test break with huggingface_hub is not the same as the tokenizers upper-bound packaging issue." + "reason": "CLI completion breakage with huggingface_hub is unrelated to tokenizers version pinning." }, { "left": "issue:43931", "right": "issue:44368", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch and a tie_word_embeddings warning are different concerns." + "reason": "Qwen3-VL weight-shape mismatch is unrelated to a tie_word_embeddings warning." }, { "left": "issue:43906", "right": "issue:46202", "accept": false, - "reason": "An isolated reproduction of another issue is not enough to match the Gemma pipeline roundtrip failure." + "reason": "Isolated reproduction of another issue is not the same as Gemma pipeline roundtrip failure on \u2581." }, { "left": "issue:45987", "right": "issue:46082", "accept": false, - "reason": "StaticCache seq_length contract bug and LlamaConfig head_dim validation are different code paths." + "reason": "StaticCache seq-length contract issue is unrelated to LlamaConfig head_dim validation." }, { - "left": "issue:44361", - "right": "issue:44749", + "left": "issue:45103", + "right": "issue:46858", "accept": false, - "reason": "Tokenizer AttributeError on tasks is unrelated to a post-upgrade filtering slowdown." + "reason": "auto_docstring annotations crash is unrelated to static-cache seq-length device-tensor arithmetic." }, { - "left": "issue:45103", - "right": "issue:46858", + "left": "issue:44361", + "right": "issue:44749", "accept": false, - "reason": "auto_docstring annotations crash and cache seq_length tensor arithmetic are unrelated regressions." + "reason": "MLukeTokenizer task AttributeError is unrelated to a training slowdown regression." }, { "left": "issue:42199", "right": "issue:46851", "accept": false, - "reason": "DETR cardinality/background-class logic does not match Qwen3.5 variable-length position_ids." + "reason": "DETR cardinality error and Qwen3.5 variable-length position_ids are different code paths." }, { "left": "issue:43994", "right": "issue:46726", "accept": false, - "reason": "SigLIP2 doc/model mismatch plus quantization failure is unrelated to missing vision-backend image processor handling." + "reason": "SigLIP2 nonsensical outputs are unrelated to AutoProcessor missing-backend handling." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "A tie_word_embeddings warning is not the same as a stale num_return_sequences warning in paged generate()." - }, - { - "left": "issue:44188", - "right": "issue:46202", - "accept": false, - "reason": "torch.compile kernel divergence is unrelated to the Gemma pipeline roundtrip tokenization issue." + "reason": "Different warning issues: tie_word_embeddings vs stale num_return_sequences warning." }, { "left": "issue:46132", "right": "issue:47145", "accept": false, - "reason": "AttentionInterface.register behavior change and FP8 weight_scale omission are distinct bugs." + "reason": "AttentionInterface.register behavior change is unrelated to FP8 checkpoint scale application." }, { "left": "issue:46752", "right": "issue:47020", "accept": false, - "reason": "Empty chat template crash and XLMRobertaTokenizer Unigram type mismatch are unrelated." + "reason": "Empty chat template list crash is unrelated to XLMRobertaTokenizer Unigram init mismatch." }, { - "left": "issue:43701", - "right": "issue:44749", + "left": "issue:44188", + "right": "issue:46202", "accept": false, - "reason": "Checkpoint key mismatch does not share the same underlying cause as the data-filtering slowdown." + "reason": "torch.compile attention-kernel branching bug is unrelated to Gemma pipeline token roundtrip failure." }, { "left": "issue:45593", "right": "issue:47068", "accept": false, - "reason": "Missing auxiliary losses and a CUDA tensor conditional synchronization issue are different." + "reason": "Missing auxiliary losses in D-FINE is unrelated to deepspeed_sp_compute_loss host synchronization." }, { "left": "issue:43976", "right": "issue:47429", "accept": false, - "reason": "Python version support breakage is unrelated to the tokenizers version upper-bound packaging problem." + "reason": "Python version compatibility regression is unrelated to tokenizers upper-bound packaging issue." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "resume_from_checkpoint key mismatch is unrelated to the training slowdown regression." }, { "left": "issue:44483", "right": "issue:46202", "accept": false, - "reason": "Chat completions request rejection is unrelated to the Gemma pipeline roundtrip failure." + "reason": "Chat completions API rejection is unrelated to Gemma pipeline roundtrip token handling." }, { "left": "issue:43784", "right": "issue:46399", "accept": false, - "reason": "Missing nn import and a missing decorator on apply_rotary_pos_emb registration are unrelated." + "reason": "Import-time NameError in sentence-transformers is unrelated to Qwen3.5 kernel registration failure." }, { "left": "issue:44188", "right": "issue:46421", "accept": false, - "reason": "These are both compile/cache-related, but they affect different mechanisms and code paths." + "reason": "Both involve compile/cache behavior, but they describe different failure modes and code paths." }, { "left": "issue:44749", "right": "issue:45656", "accept": false, - "reason": "A slowdown regression is not the same as optimizer.step being called twice under deepspeed." - }, - { - "left": "issue:44297", - "right": "issue:44623", - "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch and processor.save_pretrained missing files are distinct save-path bugs." - }, - { - "left": "issue:39692", - "right": "issue:46726", - "accept": false, - "reason": "SigLIP2 documentation example errors are unrelated to AutoProcessor backend-missing behavior." - }, - { - "left": "issue:43065", - "right": "issue:46123", - "accept": false, - "reason": "SAM3 pixel decoder dummy Conv2d and mask-generation partial-batch handling are unrelated." - }, - { - "left": "issue:43475", - "right": "issue:46821", - "accept": false, - "reason": "Missing fpn_position_embeddings in SAM3 video and TimesFM window_size AttributeError are different issues." + "reason": "Training slowdown after upgrade is unrelated to optimizer step being called twice under deepspeed." } ] }, @@ -15192,10 +15118,11 @@ "issue:28282", "issue:30990", "issue:36010", + "issue:39692", "issue:41628", "issue:42994", + "issue:43065", "issue:43232", - "issue:43278", "issue:43475", "issue:43482", "issue:43643", @@ -15207,29 +15134,29 @@ "issue:44162", "issue:44188", "issue:44279", + "issue:44297", "issue:44373", "issue:44451", "issue:44623", "issue:44661", "issue:44869", "issue:44871", - "issue:44933", "issue:44945", "issue:44964", - "issue:45081", - "issue:45103", "issue:45137", "issue:45230", - "issue:45310", "issue:45356", "issue:45375", "issue:45397", "issue:45478", "issue:45561", "issue:45656", + "issue:46123", "issue:46132", "issue:46326", "issue:46463", + "issue:46726", + "issue:46821", "issue:46851", "issue:46877", "issue:47000", @@ -15238,240 +15165,225 @@ "issue:47429" ], "soft_pairs": [ - "issue:43646|issue:44162", + "issue:44297|issue:44623", + "issue:39692|issue:46726", + "issue:43065|issue:46123", + "issue:43475|issue:46821", "issue:44964|issue:45137", "issue:43475|issue:43643", "issue:42994|issue:43482", - "issue:45230|issue:45397", + "issue:43646|issue:44162", "issue:46463|issue:46877", - "issue:44623|issue:45656", + "issue:45230|issue:45397", "issue:44373|issue:46851", + "issue:44623|issue:45656", "issue:28282|issue:30990", "issue:44871|issue:46132", "issue:43723|issue:46326", "issue:44188|issue:46851", - "issue:45561|issue:47217", "issue:46877|issue:47000", + "issue:45561|issue:47217", "issue:43232|issue:43673", "issue:44451|issue:47429", + "issue:44279|issue:47429", "issue:44038|issue:44623", "issue:44869|issue:45356", - "issue:44279|issue:47429", "issue:36010|issue:41628", "issue:45375|issue:45478", "issue:44661|issue:47059", - "issue:43873|issue:44945", - "issue:41628|issue:44933", - "issue:43278|issue:45137", - "issue:45310|issue:45375", - "issue:45081|issue:45103" + "issue:43873|issue:44945" ], - "prepared_review_unit_hash": "7bb680d17fdd04bb6c71a78282809ba21dcb41473349ce42854ec695e6047aab", + "prepared_review_unit_hash": "0f7dda074c9894dadba4425d8a8b807e5f66c5a6a395756f617e1d5f042633e6", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 34857, - "estimated_input_tokens": 8715, - "estimated_eval_tokens": 17686 + "serialized_chars": 35373, + "estimated_input_tokens": 8844, + "estimated_eval_tokens": 17944 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most items are distinct user reports across unrelated regressions, model families, or tooling issues. The only strong duplicate-adjacent cluster is the Qwen3.5-Moe v5.4.0 from_pretrained/config regression.", - "confidence": 0.83, - "canonical_issue_reason": "issue:45310 is the clearest user-facing report of the Qwen3.5-Moe transformers>=5.4.0 regression, with the affected model, version, and failure mode spelled out.", + "summary": "This cluster is a loose collection of mostly unrelated Transformers regressions and compatibility bugs, with a few broad v5-era breakages around loading, tokenizers/processors, generation, and quantization. I do not see any soft-pair duplicates that look mergeable into one fix.", + "confidence": 0.57, + "canonical_issue_reason": "issue:43646 is the best umbrella candidate: it is broad, v5-specific, and closer to the recurring theme of core model initialization regressions than the more niche or model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the best representative issue because it is concrete, broadly phrased enough to capture the regression, and directly tied to the affected model/version.", + "best_issue_reason": "issue:43646 is the most representative single issue in the set because it captures a central Transformers 5.0 breakage rather than a narrow model- or docs-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43646", - "right": "issue:44162", + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both involve save_pretrained behavior, but one is tokenizer config naming and the other is processor file emission; different code paths and fixes." + }, + { + "left": "issue:39692", + "right": "issue:46726", "accept": false, - "reason": "Both are transformer-related, but they describe different problems: custom model init vs ESM2 breakage." + "reason": "SigLIP2 documentation example errors and AutoProcessor backend-missing errors are different problems, one in docs/example usage and one in runtime backend detection." + }, + { + "left": "issue:43065", + "right": "issue:46123", + "accept": false, + "reason": "SAM3 pixel decoder dummy layer behavior is unrelated to MaskGenerationPipeline dropping the last partial batch." + }, + { + "left": "issue:43475", + "right": "issue:46821", + "accept": false, + "reason": "A missing Sam3VisionEncoderOutput attribute and a TimesFM window_size AttributeError are unrelated model-specific bugs." }, { "left": "issue:44964", "right": "issue:45137", "accept": false, - "reason": "A model-loading regression and a DeepSpeed ZeRO3 deque error are unrelated code paths." + "reason": "Phi-4 multimodal loading failure and DeepSpeed ZeRO3 deque pop are distinct failures in different subsystems." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "One is a SAM3 vision encoder attribute bug; the other is AutoConfig/trust_remote_code returning incomplete fields." + "reason": "Missing Sam3VisionEncoderOutput fields and AutoConfig trust_remote_code missing fields are not the same underlying bug." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized model saving and GGUF loading with v5 are different failures." + "reason": "Quantized model saving failure and Qwen2.5-GGUF loading failure are separate quantization/load-path issues." }, { - "left": "issue:45230", - "right": "issue:45397", + "left": "issue:43646", + "right": "issue:44162", "accept": false, - "reason": "A generic bug report and a gemma-4 Zero3 from_pretrained issue do not describe the same bug." + "reason": "Both are broad breakages, but one is custom model initialization on v5 and the other is an ESM2-specific regression; not the same fix." }, { "left": "issue:46463", "right": "issue:46877", "accept": false, - "reason": "Decoder input validation error messaging and auto_docstring stdout logging are unrelated." + "reason": "Incorrect T5 decoder validation messaging and auto_docstring stdout diagnostics are unrelated; one is model validation, the other is tooling noise." }, { - "left": "issue:44623", - "right": "issue:45656", + "left": "issue:45230", + "right": "issue:45397", "accept": false, - "reason": "processor.save_pretrained missing files is not the same as an optimizer stepping twice under deepspeed." + "reason": "A placeholder bug report is not the same as a Gemma-4 ZeRO3 from_pretrained regression." }, { "left": "issue:44373", "right": "issue:46851", "accept": false, - "reason": "Wrong docstring for position_ids is documentation-only; the other is a runtime position_ids bug." + "reason": "Wrong docstring text for position_ids is documentation-only, while Qwen3.5 variable-length input failure is a runtime logic bug." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor save_pretrained file emission and deepspeed optimizer double-step behavior are unrelated." }, { "left": "issue:28282", "right": "issue:30990", "accept": false, - "reason": "PyTorch-missing ImportError and Sentence Transformers loading hang are different symptoms with different causes." + "reason": "PyTorch missing import error and Sentence Transformers loading hang have different symptoms and likely different root causes." }, { "left": "issue:44871", "right": "issue:46132", "accept": false, - "reason": "Gemma eos_token_id config inconsistency and AttentionInterface.register behavior change are unrelated." + "reason": "Gemma eos_token_id config mismatch and AttentionInterface.register behavior changes are not the same code-path problem." }, { "left": "issue:43723", "right": "issue:46326", "accept": false, - "reason": "Tokenizer loading in v5 and continue_final_message with mistral_common are different tokenizer bugs." + "reason": "Tokenizer loading in v5 and continue_final_message with mistral_common tokenizers are separate tokenizer issues." }, { "left": "issue:44188", "right": "issue:46851", "accept": false, - "reason": "torch.compile attention kernel branching and Qwen3.5 variable-length position_ids are different issues." + "reason": "torch.compile branching divergence in attention kernels is unrelated to Qwen3.5 variable-length position_ids handling." }, { - "left": "issue:45561", - "right": "issue:47217", + "left": "issue:46877", + "right": "issue:47000", "accept": false, - "reason": "A test xdist race on captured_info.txt is unrelated to VL image-placeholder handling." + "reason": "Import-time doc-lint stdout noise is unrelated to RTDetrLoss infeasible cost-matrix crashes." }, { - "left": "issue:46877", - "right": "issue:47000", + "left": "issue:45561", + "right": "issue:47217", "accept": false, - "reason": "Import-time docstring diagnostics and RTDetrLoss cost-matrix infeasibility are unrelated." + "reason": "pytest-xdist captured_info.txt race and VL image-placeholder handling are completely different bugs." }, { "left": "issue:43232", "right": "issue:43673", "accept": false, - "reason": "Both are generation-related, but one concerns sync_gpus kwargs propagation and the other missing GenerationMixin cache." + "reason": "Generation kwargs update after sync_gpus and missing GenerationMixin cache in chunked_prefill are different generation-path regressions." }, { "left": "issue:44451", "right": "issue:47429", "accept": false, - "reason": "ScandiBERT loading failure and a tokenizers upper-bound packaging issue are not the same bug." + "reason": "Loading a specific model and an upper-bound packaging constraint on tokenizers are not the same underlying issue." + }, + { + "left": "issue:44279", + "right": "issue:47429", + "accept": false, + "reason": "A generic dependency issue and the tokenizers version bound problem are too different to merge as one fix." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Qwen3-VL-Moe breakage and processor.save_pretrained missing files affect different paths." + "reason": "Qwen3-VL-Moe/v5 incompatibility and processor save_pretrained missing files are unrelated." }, { "left": "issue:44869", "right": "issue:45356", "accept": false, - "reason": "Whisper timestamp decoding crash and Kimi-K2.5 tokenizer regression are different tokenizer edge cases." - }, - { - "left": "issue:44279", - "right": "issue:47429", - "accept": false, - "reason": "A generic dependency issue is not the same as the specific tokenizers version-cap problem." + "reason": "Whisper word timestamp decoding crash and Kimi-K2.5 tokenizer regression affect different tokenization/decoding paths." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they concern different symbols and different API breakages." + "reason": "GenerationMixin import failure and AutoImageProcessor import failure are both import errors, but for different symbols and subsystems." }, { "left": "issue:45375", "right": "issue:45478", - "accept": true, - "reason": "Both reports point to the same Qwen3.5-Moe transformers>=5.4.0 from_pretrained regression; one just phrases it via the missing config field." + "accept": false, + "reason": "Missing deepstack_visual_indexes in a config class and a Qwen3.5 Moe from_pretrained error are related by model family only, not by a shared defect." }, { "left": "issue:44661", "right": "issue:47059", "accept": false, - "reason": "add-new-model-like/tokenizer mapping logic and a cli completion attribute error are unrelated." - }, - { - "left": "issue:43873", - "right": "issue:44945", - "accept": false, - "reason": "Quantization offloading behavior and incorrect pipeline-parallel LLM output are different failure modes." - }, - { - "left": "issue:41628", - "right": "issue:44933", - "accept": false, - "reason": "Missing AutoImageProcessor import and a nonexistent image_utils import are similar API complaints but not the same concrete bug." - }, - { - "left": "issue:43278", - "right": "issue:45137", - "accept": false, - "reason": "BF16-to-FP32 dtype drift in evaluation is unrelated to a Zero3 deque pop error." - }, - { - "left": "issue:45310", - "right": "issue:45375", - "accept": true, - "reason": "Same underlying Qwen3.5-Moe v5.4.0 regression: from_pretrained fails because required config data is missing or mishandled." - }, - { - "left": "issue:45081", - "right": "issue:45103", - "accept": false, - "reason": "Mistral tokenizer regex patching and auto_docstring future-annotations crashes are unrelated." - }, - { - "left": "issue:45375", - "right": "issue:45478", - "accept": true, - "reason": "These are duplicate descriptions of the same Qwen3.5-Moe from_pretrained/config regression in transformers>=5.4.0." + "reason": "add-new-model-like tokenizer mapping failure and huggingface_hub CLI completion breakage are unrelated integration bugs." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The only clearly duplicate-adjacent cluster is the Qwen3.5-Moe v5.4.0 from_pretrained/config regression, and the true soft-edge verdicts there are supported by the titles. The false verdicts are cautious and do not overstate sameness." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -15496,17 +15408,15 @@ "issue:41628", "issue:42175", "issue:42371", - "issue:42673", "issue:42831", "issue:42898", "issue:43232", + "issue:43278", "issue:43479", - "issue:43704", "issue:43784", "issue:43824", "issue:43976", "issue:44038", - "issue:44060", "issue:44186", "issue:44222", "issue:44351", @@ -15518,9 +15428,12 @@ "issue:44857", "issue:44933", "issue:44945", + "issue:45081", "issue:45092", - "issue:45125", + "issue:45103", + "issue:45137", "issue:45310", + "issue:45375", "issue:45405", "issue:45406", "issue:45561", @@ -15529,7 +15442,6 @@ "issue:45753", "issue:45812", "issue:46202", - "issue:46693", "issue:46701", "issue:46829", "issue:46877", @@ -15538,8 +15450,12 @@ "issue:47085" ], "soft_pairs": [ - "issue:42371|issue:45092", + "issue:41628|issue:44933", + "issue:43278|issue:45137", + "issue:45310|issue:45375", + "issue:45081|issue:45103", "issue:29925|issue:30333", + "issue:42371|issue:45092", "issue:44857|issue:46877", "issue:45406|issue:46202", "issue:46202|issue:46701", @@ -15553,10 +15469,10 @@ "issue:30333|issue:33290", "issue:45405|issue:46877", "issue:44743|issue:47068", - "issue:45561|issue:47059", "issue:42175|issue:45712", "issue:42898|issue:44568", "issue:44945|issue:45753", + "issue:45561|issue:47059", "issue:42831|issue:44222", "issue:42898|issue:45812", "issue:44496|issue:45561", @@ -15565,235 +15481,235 @@ "issue:44843|issue:47085", "issue:44568|issue:47085", "issue:43479|issue:46829", - "issue:44186|issue:47085", - "issue:42673|issue:46693", - "issue:44060|issue:45125", - "issue:42673|issue:43704" + "issue:44186|issue:47085" ], - "prepared_review_unit_hash": "d3dcb4d4813b67a3082306a9a2f68188665d1dc2f2515dd73ce2e684f6d47774", + "prepared_review_unit_hash": "b4c3822ff0dc5b14464dcf8bc8907ecf703bd0339cd2f1bf41288be1a3b73664", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36920, - "estimated_input_tokens": 9230, - "estimated_eval_tokens": 18716 + "soft_pair_count": 32, + "serialized_chars": 37373, + "estimated_input_tokens": 9344, + "estimated_eval_tokens": 18944 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous issue reports across unrelated bugs. The only clear duplicate soft edge is the repeated Qwen3ForCausalLM VRAM leak report (#42673 / #43704).", - "confidence": 0.93, - "canonical_issue_reason": "#42673 is the strongest anchor in the only true duplicate pair: #43704 repeats the same VRAM-leak title, while #42673 has more discussion/context.", + "summary": "This looks like a mostly heterogeneous set of Transformers issues with only superficial lexical similarity; I did not find any soft-edge pair that clearly represents the same underlying bug/change, so no duplicates were accepted.", + "confidence": 0.88, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "#42673 is the best issue to keep as the representative ticket because it is the earliest, has more context, and matches an exact duplicate (#43704).", + "best_issue_reason": "Issue 44843 is the strongest standalone bug report here: it has a specific failing code path, clear offline-mode impact, and concrete reproducer details. It is still not a good canonical for the whole cluster because the rest are largely unrelated.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:42371", - "right": "issue:45092", + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Different import failures in different modules; no shared code-path or concrete fix target." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Unrelated bugs: embedding dtype regression vs DeepSpeed ZeRO3 deque error." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Same model family, but one is a generic from_pretrained failure and the other is a specific config-field bug; not clearly the same defect." + }, + { + "left": "issue:45081", + "right": "issue:45103", "accept": false, - "reason": "Different problems: TF32 backend settings vs. old InternVL2 meta-initialization incompatibility." + "reason": "Both are crashes, but in different utilities and different failure modes; not the same underlying bug." }, { "left": "issue:29925", "right": "issue:30333", "accept": false, - "reason": "Unrelated: SigLIP tokenizer fast-tokenizer enforcement vs. MLflow reporting after failed training." + "reason": "Completely unrelated: tokenizer fast-mode enforcement vs MLFlow reporting after failed training." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 configuration issue and meta-init incompatibility are different problems." }, { "left": "issue:44857", "right": "issue:46877", "accept": false, - "reason": "Different bugs in different areas: LwDetrImageLoss AMP/CUDA crash vs. docstring lint output at import time." + "reason": "Training-loss crash and docstring import-time logging are unrelated." }, { "left": "issue:45406", "right": "issue:46202", "accept": false, - "reason": "Both mention Gemma, but one is a missing processor attribute and the other is a tokenizer roundtrip failure." + "reason": "Both involve Gemma, but one is a processor attribute bug and the other is a tokenizer roundtrip issue." }, { "left": "issue:46202", "right": "issue:46701", "accept": false, - "reason": "Same model family, but distinct failures and code paths: tokenizer roundtrip vs. image-text pipeline not working." + "reason": "Different pipeline failures with different symptoms and code paths." }, { "left": "issue:45593", "right": "issue:46877", "accept": false, - "reason": "Unrelated: D-FINE auxiliary-loss behavior vs. auto_docstring printing diagnostics at import." + "reason": "D-FINE loss behavior vs docstring diagnostics; no overlap in underlying bug." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Generation-sync kwargs issue vs. a janus image generation test failure; not the same bug." + "reason": "Generation kwargs sync bug and Janus image-generation test failure are distinct." }, { "left": "issue:43784", "right": "issue:43824", "accept": false, - "reason": "Different import errors affecting different symbols and models." + "reason": "Both are import errors, but for different symbols and unrelated causes." }, { "left": "issue:42175", "right": "issue:44933", "accept": false, - "reason": "Backend installation dependency issue vs. a missing image_utils import; no shared code path." + "reason": "Packaging/backend dependency issue vs missing image_utils import; unrelated." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Related model family, but different model variants and errors across different versions." + "reason": "Same broad Qwen area, but different model variants and failure reports; not enough to treat as one concrete bug." }, { "left": "issue:41628", "right": "issue:44351", "accept": false, - "reason": "Different missing imports: AutoImageProcessor vs. HybridCache." + "reason": "Different missing exports from transformers; no shared defect shown." }, { "left": "issue:44222", "right": "issue:45310", "accept": false, - "reason": "FP8 save_pretrained/moe behavior is not the same as Qwen3.5 MoE from_pretrained failure." + "reason": "FP8 save_pretrained behavior and Qwen3.5 loading error are different bugs." }, { "left": "issue:30333", "right": "issue:33290", "accept": false, - "reason": "Different runtime issues: MLflow reporting vs. DeepSpeed OOM with Adafactor." + "reason": "MLFlow reporting vs deepspeed OOM are unrelated." }, { "left": "issue:45405", "right": "issue:46877", "accept": false, - "reason": "PEFT version pinning issue vs. import-time docstring diagnostics; unrelated." + "reason": "Dependency version pinning vs docstring import logging are unrelated." }, { "left": "issue:44743", "right": "issue:47068", "accept": false, - "reason": "Different generation/runtime issues in separate code paths." - }, - { - "left": "issue:45561", - "right": "issue:47059", - "accept": false, - "reason": "Pytest-xdist file race vs. CLI breakage with huggingface_hub; unrelated." + "reason": "Recurrent-state reset bug and CUDA host-sync conditional are different generation issues." }, { "left": "issue:42175", "right": "issue:45712", "accept": false, - "reason": "Dependency/backends packaging issue vs. dummy class cleanup in repo checks." + "reason": "Dependency packaging issue vs leftover dummy classes leaking into dir(); unrelated." }, { "left": "issue:42898", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer-related v5 regressions, but they describe different behaviors and likely different fixes." + "reason": "Both tokenizer regressions in v5, but they affect different APIs and different models." }, { "left": "issue:44945", "right": "issue:45753", "accept": false, - "reason": "Pipeline-parallelism output bug vs. an image-triggered infinite loop in Qwen3_5." + "reason": "Pipeline-parallel output corruption and a specific image infinite loop are not the same bug." + }, + { + "left": "issue:45561", + "right": "issue:47059", + "accept": false, + "reason": "Test xdist file race and CLI completion attribute error are unrelated test/runtime issues." }, { "left": "issue:42831", "right": "issue:44222", "accept": false, - "reason": "Both mention FP8, but one is an accuracy issue and the other is save_pretrained/moe behavior." + "reason": "Both mention FP8, but one is accuracy and the other is save_pretrained behavior; different issues." }, { "left": "issue:42898", "right": "issue:45812", "accept": false, - "reason": "Different tokenizer regressions: cleanup-space behavior vs. wrong token IDs for Granite models." + "reason": "Different tokenizer regressions: cleanup-space behavior vs wrong token IDs." }, { "left": "issue:44496", "right": "issue:45561", "accept": false, - "reason": "Unrelated: unrecognized model/config issue vs. pytest-xdist captured_info race." + "reason": "Model loading/config error vs xdist race in testing utils; unrelated." }, { "left": "issue:30990", "right": "issue:35141", "accept": false, - "reason": "Loading hang vs. embedding reinitialization bug; not the same underlying problem." + "reason": "Loading stalls and embedding reinitialization are different model/setup bugs." }, { "left": "issue:43976", "right": "issue:44933", "accept": false, - "reason": "Python-version compatibility issue vs. missing import from image_utils." + "reason": "Python-version compatibility issue vs missing import symbol; unrelated." }, { "left": "issue:44843", "right": "issue:47085", "accept": false, - "reason": "Offline model_info call bug vs. tokenizer thread-safety regression; different code paths." + "reason": "Offline-mode Mistral regex patch bug and tokenizer thread-safety regression are different code paths." }, { "left": "issue:44568", "right": "issue:47085", "accept": false, - "reason": "Tokenizer special-token regression vs. fast tokenizer thread-safety; not duplicates." + "reason": "Tokenizer special-token regression and tokenizer thread-safety regression are not the same defect." }, { "left": "issue:43479", "right": "issue:46829", "accept": false, - "reason": "Config initialization bug vs. output object field mismatch; unrelated." + "reason": "Multimodal config defaulting bug vs incorrect hidden_states output in classification are unrelated." }, { "left": "issue:44186", "right": "issue:47085", "accept": false, - "reason": "LayoutLMv2 tokenizer crash on NER/padding vs. tokenizer thread-safety; different failures." - }, - { - "left": "issue:42673", - "right": "issue:46693", - "accept": false, - "reason": "Both are performance/runtime-related, but one is a VRAM leak in multi-dataloader threads and the other is a long-context FA2 regression." - }, - { - "left": "issue:44060", - "right": "issue:45125", - "accept": false, - "reason": "Different Qwen3 model bugs: tied-weights warning vs. missing tensor-parallel plan." - }, - { - "left": "issue:42673", - "right": "issue:43704", - "accept": true, - "reason": "Exact duplicate: same title and same Qwen3ForCausalLM VRAM leak in multiple dataloader threads." + "reason": "LayoutLMv2Tokenizer crash and fast-tokenizer thread-safety regression are different bugs." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet: the cluster is heterogeneous, and the only clearly supported duplicate is #42673/#43704, which share the same title and bug description. The representative-ticket rationale is conservative and backed by earlier creation date plus more discussion/context. The soft-edge rejections also stay cautious and do not overclaim merges or duplicates beyond what the packet supports." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -15819,15 +15735,15 @@ "issue:42222", "issue:42673", "issue:42796", - "issue:42831", "issue:42898", "issue:43116", "issue:43295", "issue:43479", + "issue:43704", "issue:43723", "issue:43874", "issue:44016", - "issue:44075", + "issue:44060", "issue:44079", "issue:44190", "issue:44222", @@ -15841,8 +15757,8 @@ "issue:44869", "issue:45092", "issue:45103", + "issue:45125", "issue:45200", - "issue:45245", "issue:45310", "issue:45405", "issue:45600", @@ -15854,21 +15770,24 @@ "issue:46209", "issue:46231", "issue:46431", - "issue:46665", + "issue:46693", "issue:46909", "issue:47059", "issue:47085" ], "soft_pairs": [ + "issue:44060|issue:45125", + "issue:42673|issue:46693", + "issue:42673|issue:43704", "issue:44623|issue:45310", "issue:39692|issue:46909", "issue:43723|issue:44568", + "issue:30333|issue:34689", "issue:30333|issue:36331", "issue:43479|issue:44079", - "issue:30333|issue:34689", "issue:45600|issue:47059", - "issue:42898|issue:44448", "issue:42898|issue:43295", + "issue:42898|issue:44448", "issue:42673|issue:47085", "issue:42898|issue:44779", "issue:43479|issue:44869", @@ -15884,235 +15803,214 @@ "issue:45405|issue:45712", "issue:43116|issue:44190", "issue:44448|issue:46132", - "issue:44222|issue:44297", - "issue:43479|issue:46665", - "issue:44297|issue:45656", - "issue:42831|issue:44623", - "issue:44075|issue:46231", - "issue:45245|issue:46132", - "issue:44222|issue:44623" + "issue:44222|issue:44297" ], - "prepared_review_unit_hash": "4813c4ec4f49468f42086dda4c163ba569f117dc3f9823cf919e72f3beebb619", + "prepared_review_unit_hash": "94dc4ae68a79adf7e6d1d0a267aae48681bd3e6dda1f0d9df38c119d987b34d2", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37074, - "estimated_input_tokens": 9269, - "estimated_eval_tokens": 18794 + "soft_pair_count": 28, + "serialized_chars": 36284, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly false positives: this cluster is a grab bag of unrelated Transformers bugs (tokenizers, model loading, training, docs, serving, infra). None of the soft pairs look like the same underlying issue/change, so there is no meaningful duplicate canonicalization here.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue fits this cluster; the items span many unrelated code paths and only share broad repository-level similarity.", + "summary": "The cluster contains one clear duplicate issue pair: issue:42673 and issue:43704 have the exact same title and describe the same Qwen3ForCausalLM VRAM leak when used in multiple dataloader threads. The rest of the soft pairs are thematic overlaps rather than the same underlying bug.", + "confidence": 0.97, + "canonical_issue_reason": "issue:42673 is the best canonical issue because it is the earlier report of the exact same VRAM-leak bug described again in issue:43704.", "canonical_pr_reason": null, - "best_issue_reason": "No strong best issue for deduping. The closest repeated theme is tokenizer/v5 regressions, but even those pairs still describe different bugs and fixes.", + "best_issue_reason": "issue:42673 is the strongest representative issue in the cluster because it matches issue:43704 exactly in title and symptom, and it was created earlier.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Both are Qwen-family model issues, but they describe different failures: tied-weight warnings vs a missing tensor-parallel plan." + }, + { + "left": "issue:42673", + "right": "issue:46693", + "accept": false, + "reason": "Both concern resource/performance behavior, but one is a VRAM leak in dataloader threads and the other is a long-context FA2 regression. Different bugs." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Exact same title and same symptom: Qwen3ForCausalLM leaks VRAM when used in multiple dataloader threads. Clear duplicate." + }, { "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "Different failures: processor.save_pretrained missing files vs a Qwen3.5 MoE from_pretrained error. Same broad loading area, different code path." + "reason": "One is about processor.save_pretrained missing files; the other is a from_pretrained error for Qwen3.5 MoE. Related area, not the same bug." }, { "left": "issue:39692", "right": "issue:46909", "accept": false, - "reason": "Both are broken docs examples, but for different models and different mistakes; not the same underlying bug." + "reason": "Both are documentation/example failures, but they involve different examples and different errors." }, { "left": "issue:43723", "right": "issue:44568", "accept": false, - "reason": "AutoTokenizer loading in v5 and special_tokens/BOS-EOS behavior are distinct tokenizer bugs." + "reason": "Both are tokenizer regressions, but one is an AutoTokenizer loading issue in v5 and the other is missing BOS/EOS addition for a specific tokenizer." }, { "left": "issue:30333", - "right": "issue:36331", + "right": "issue:34689", "accept": false, - "reason": "MLFlow reporting on failed training jobs is unrelated to Trainer.compute_loss() receiving num_items_in_batch." + "reason": "Different subsystems and failure modes: MLflow reporting for failed training vs a model-loading regression for Llama 3.2 Vision." }, { - "left": "issue:43479", - "right": "issue:44079", + "left": "issue:30333", + "right": "issue:36331", "accept": false, - "reason": "Phi4 config initialization and ModelOutput key assignment are separate internals with no shared failure mode." + "reason": "Unrelated issues: MLflow job reporting vs a CustomTrainer.compute_loss signature change." }, { - "left": "issue:30333", - "right": "issue:34689", + "left": "issue:43479", + "right": "issue:44079", "accept": false, - "reason": "MLFlow job-state reporting is unrelated to Llama 3.2 Vision model loading breakage in 4.46.2." + "reason": "Different bugs: multimodal config default initialization vs ModelOutput key assignment when a field was previously None." }, { "left": "issue:45600", "right": "issue:47059", "accept": false, - "reason": "auto_mappings.py stale config references and huggingface_hub CLI breakage are different CI/runtime issues." + "reason": "Different areas: stale auto_mappings config references vs a CLI failure with huggingface_hub 1.22.0." }, { "left": "issue:42898", - "right": "issue:44448", + "right": "issue:43295", "accept": false, - "reason": "Tokenization cleanup behavior in v5 is not the same as Pegasus output differences between v4 and v5." + "reason": "Both are v5 regressions, but one is about clean_up_tokenization_spaces behavior and the other is processor.tokenizer access/images-to-tokenizer behavior." }, { "left": "issue:42898", - "right": "issue:43295", + "right": "issue:44448", "accept": false, - "reason": "clean_up_tokenization_spaces regression and processor.tokenizer/images handling regression affect different APIs." + "reason": "One is a tokenizer cleanup regression, the other is a Pegasus v4/v5 output difference. Not the same bug." }, { "left": "issue:42673", "right": "issue:47085", "accept": false, - "reason": "VRAM leakage across dataloader threads is not the same as fast tokenizer thread-unsafety." + "reason": "Both involve threading/memory concerns, but one is a Qwen VRAM leak and the other is tokenizer thread-safety. Different underlying problem." }, { "left": "issue:42898", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer-related v5 regressions, but they concern different tokenizers and different broken behavior." + "reason": "Both are tokenizer-related, but Deepseek incorrect results and clean_up_tokenization_spaces behavior changes are distinct failures." }, { "left": "issue:43479", "right": "issue:44869", "accept": false, - "reason": "Phi4MultimodalConfig default init bug is unrelated to Whisper word timestamp decode crashing on trailing replacement characters." + "reason": "Completely different features: multimodal config defaults vs Whisper word-timestamp decoding crash." }, { "left": "issue:44016", "right": "issue:46231", "accept": false, - "reason": "A notebook syntax error is not the same as a docs parameter-name mismatch." + "reason": "Different documentation problems: a notebook syntax error vs a parameter-name mismatch in set_training docs." }, { "left": "issue:22355", "right": "issue:28282", "accept": false, - "reason": "Missing transformers.onnx module and missing PyTorch for AutoModel are separate import/dependency issues." + "reason": "Both are import-related, but one is missing transformers.onnx and the other is missing PyTorch for AutoModel." }, { "left": "issue:46077", "right": "issue:46123", "accept": false, - "reason": "Bidirectional attention mask construction and final partial-batch dropping are unrelated pipeline/model bugs." + "reason": "Different code paths: Gemma4 attention-mask construction vs MaskGenerationPipeline final-batch handling." }, { "left": "issue:42222", "right": "issue:45200", "accept": false, - "reason": "Broken vitpose models and Gemma 4 mm_token_type_ids defaults involve different model families and fixes." + "reason": "Different multimodal model bugs: broken vitpose models vs Gemma 4 mm_token_type_ids defaulting." }, { "left": "issue:46209", "right": "issue:46431", "accept": false, - "reason": "FP8 inference on B200 and apply_chat_template requiring a chat template are unrelated features." + "reason": "Different runtime failures: FP8 inference on B200 vs apply_chat_template on a processor without a chat template." }, { "left": "issue:44295", "right": "issue:45656", "accept": false, - "reason": "Reading a registered position_ids buffer is unrelated to DeepSpeed causing optimizer.step() to run twice." + "reason": "Different failures: reading a position_ids buffer vs optimizer.step being called twice with deepspeed." }, { "left": "issue:44734", "right": "issue:45103", "accept": false, - "reason": "Serving KV-cache continuation indexing and auto_docstring future-annotations handling are different subsystems." + "reason": "One is a transformers serve KV-cache indexing crash; the other is an auto_docstring AttributeError with future annotations." }, { "left": "issue:42796", "right": "issue:43723", "accept": false, - "reason": "Tokenizer BPE marker handling and AutoTokenizer.from_pretrained loading are different tokenizer issues." + "reason": "Both are tokenizer-related, but one is BPE-marker handling and the other is AutoTokenizer.from_pretrained loading in v5." }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Missing image-processor method and remote-code meta-init incompatibility are both multimodal loading problems, but not the same bug." + "reason": "Different model-specific issues: missing image-patches method in GLM46V processor vs remote-code incompatibility for old InternVL2 checkpoints." }, { "left": "issue:45405", "right": "issue:45712", "accept": false, - "reason": "A version-pin bump to unreleased peft and leftover dummy classes failing repo checks are unrelated maintenance issues." + "reason": "Unrelated maintenance issues: PEFT version pin bump vs leftover dummy classes leaking into repo checks." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Multi-label classification example output and local dataset loading in a different example script are unrelated." + "reason": "Different example-script failures: empty multi-label classification results vs inability to load a local dataset." }, { "left": "issue:44448", "right": "issue:46132", "accept": false, - "reason": "Pegasus v4/v5 output drift and AttentionInterface.register behavior changes do not share a concrete code path." + "reason": "Different behavioral changes: Pegasus v4/v5 output mismatch vs AttentionInterface.register altering registered function behavior." }, { "left": "issue:44222", "right": "issue:44297", "accept": false, - "reason": "FP8 save_pretrained for MoE and tokenizer_class mismatch in tokenizer_config.json are different save_pretrained bugs." - }, - { - "left": "issue:43479", - "right": "issue:46665", - "accept": false, - "reason": "Incorrect default multimodal config init is unrelated to from_pretrained behavior inside Python classes." - }, - { - "left": "issue:44297", - "right": "issue:45656", - "accept": false, - "reason": "Tokenizer config class mismatch and doubled optimizer steps are completely separate issues." - }, - { - "left": "issue:42831", - "right": "issue:44623", - "accept": false, - "reason": "FP8 accuracy problems and processor.save_pretrained missing files are different failures and fixes." - }, - { - "left": "issue:44075", - "right": "issue:46231", - "accept": false, - "reason": "SGD args being ignored is not the same as a documentation parameter-name typo." - }, - { - "left": "issue:45245", - "right": "issue:46132", - "accept": false, - "reason": "Category-count limits and AttentionInterface.register behavior changes are unrelated." - }, - { - "left": "issue:44222", - "right": "issue:44623", - "accept": false, - "reason": "MoE FP8 save_pretrained and processor.save_pretrained missing files are different APIs and defects." + "reason": "Different save_pretrained bugs: FP8 MoE save issues vs tokenizer_class mismatch in tokenizer_config.json." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, - "accepted_nontrivial_soft_edge": false, + "evaluator_result": { + "accept": false, + "feedback": "The duplicate-pair conclusion for issue:42673 and issue:43704 is grounded, but the canonical/best-issue rationale overstates the evidence. The packet supports: exact same title, same symptom, and earlier creation time for issue:42673. It does not support \"more discussion/inbound references\" because inbound references are equal (0 vs 0), and \"more complete report\" is not directly evidenced in the packet. The soft-edge verdicts themselves are mostly conservative, but the summary should avoid those unsupported claims." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -16133,22 +16031,21 @@ "issue:38175", "issue:41628", "issue:42757", + "issue:42831", "issue:42907", "issue:43122", "issue:43232", "issue:43408", "issue:43475", + "issue:43479", "issue:43504", "issue:43723", - "issue:43986", + "issue:44075", "issue:44190", "issue:44222", - "issue:44263", - "issue:44279", "issue:44295", "issue:44297", "issue:44327", - "issue:44360", "issue:44373", "issue:44484", "issue:44485", @@ -16157,253 +16054,260 @@ "issue:44977", "issue:45071", "issue:45092", + "issue:45245", "issue:45310", "issue:45357", - "issue:45381", + "issue:45656", "issue:45712", "issue:45834", "issue:46050", "issue:46082", + "issue:46132", "issue:46209", + "issue:46231", "issue:46305", - "issue:46328", "issue:46369", "issue:46402", "issue:46424", "issue:46614", - "issue:46726", - "issue:46762", + "issue:46665", "issue:46909", "issue:46932", "issue:47059", "issue:47276" ], "soft_pairs": [ + "issue:43479|issue:46665", + "issue:44297|issue:45656", + "issue:42831|issue:44623", + "issue:44075|issue:46231", "issue:44933|issue:45712", - "issue:43723|issue:44977", + "issue:45245|issue:46132", + "issue:44222|issue:44623", "issue:44373|issue:44485", - "issue:44297|issue:45310", + "issue:43723|issue:44977", "issue:43232|issue:47276", + "issue:44297|issue:45310", "issue:43504|issue:47059", - "issue:43122|issue:46209", - "issue:46614|issue:46932", "issue:44327|issue:47059", + "issue:46614|issue:46932", + "issue:43122|issue:46209", "issue:43475|issue:46305", - "issue:43232|issue:45834", "issue:41628|issue:42757", + "issue:43232|issue:45834", "issue:30990|issue:38175", "issue:45071|issue:46082", - "issue:44295|issue:44623", "issue:45092|issue:45357", - "issue:46369|issue:46909", + "issue:44295|issue:44623", "issue:44190|issue:46082", - "issue:43232|issue:46424", "issue:44222|issue:44484", + "issue:46369|issue:46909", + "issue:43232|issue:46424", "issue:42907|issue:46050", - "issue:43408|issue:46402", - "issue:44360|issue:46762", - "issue:43986|issue:46726", - "issue:45381|issue:46305", - "issue:44263|issue:45092", - "issue:44263|issue:44373", - "issue:44279|issue:46328" + "issue:43408|issue:46402" ], - "prepared_review_unit_hash": "a69c1bb91fb50af662483e0c85b27e9d38d06e69e6f822f6bb83e43aeea8ea68", + "prepared_review_unit_hash": "886ed0b27c8cf8c3a741a04261c0a25107a00c45ab1ca9c8f8c4671fb47e089b", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 35791, - "estimated_input_tokens": 8948, - "estimated_eval_tokens": 18152 + "serialized_chars": 35733, + "estimated_input_tokens": 8934, + "estimated_eval_tokens": 18124 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of Transformers regression reports around loading/saving/imports/config validation, not a single duplicate bug. None of the soft pairs look like the same underlying issue closely enough to merge.", - "confidence": 0.34, - "canonical_issue_reason": "Issue #43723 is the broadest v5 loading-related regression in the set, so it is the most reasonable anchor if one must pick a representative issue; however, the overall cluster is too heterogeneous for a strong canonical choice.", + "summary": "These items are mostly unrelated issues spanning tokenization, config validation, serialization, FP8, SAM3, and dependency-compatibility paths. None of the soft pairs look like true duplicates of the same underlying bug or change.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue: the cluster is heterogeneous and the soft links do not indicate a shared underlying bug.", "canonical_pr_reason": null, - "best_issue_reason": "#43723 is the best single issue anchor because it is general and user-facing, but it still only covers one narrow loading problem and does not truly represent the rest of the mixed cluster.", + "best_issue_reason": "issue:44297 is the clearest representative bug report: it is concrete, reproducible, and about a specific save_pretrained tokenizer metadata mismatch.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43479", + "right": "issue:46665", + "accept": false, + "reason": "Different problems: Phi4MultimodalConfig default initialization vs from_pretrained behavior inside Python classes." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save metadata mismatch and optimizer stepping twice in deepspeed are unrelated code paths." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy regression and missing processor save files are different serialization/compute issues." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "SGD args being ignored is a runtime bug; the other is only a doc parameter-name mismatch." + }, { "left": "issue:44933", "right": "issue:45712", "accept": false, - "reason": "Both mention import-related breakage, but one is a missing import from image_utils and the other is leftover dummy classes leaking into dir(transformers); different root causes and fixes." + "reason": "Both mention imports, but one is a missing image_utils export and the other is cleanup of dummy classes." }, { - "left": "issue:43723", - "right": "issue:44977", + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Category-count limit error and AttentionInterface.register behavior are unrelated." + }, + { + "left": "issue:44222", + "right": "issue:44623", "accept": false, - "reason": "Tokenizer loading failure and Qwen3.5 flash-attention generation failure are unrelated bugs affecting different code paths." + "reason": "Both involve save_pretrained, but one is FP8 MOE saving and the other is processor file omission." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "One is a wrong docstring for position_ids; the other is a RoPE implementation discussion. Not the same bug." + "reason": "Docstring typo for position_ids is unrelated to max_shard_size behavior in save_pretrained." }, { - "left": "issue:44297", - "right": "issue:45310", + "left": "issue:43723", + "right": "issue:44977", "accept": false, - "reason": "Both are save/load-related, but one is tokenizer_class metadata on save and the other is a Qwen3.5 MoE from_pretrained regression; different failures." + "reason": "Tokenizer loading in v5 and Qwen3.5 flash-attention generation are different failure modes." }, { "left": "issue:43232", "right": "issue:47276", "accept": false, - "reason": "Both touch synced_gpus/generation warnings, but one is a model-kwarg update bug and the other is a misleading warning in continuous batching; not the same underlying defect." + "reason": "Both mention synced_gpus, but one is generation state update logic and the other is a warning message." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer_class mismatch and Qwen3.5 MoE from_pretrained error are different bugs." }, { "left": "issue:43504", "right": "issue:47059", "accept": false, - "reason": "A Beit pretrained preset loading bug and a CLI completion failure from huggingface_hub are unrelated." + "reason": "A pretrained model legacy-field loading bug is unrelated to a huggingface_hub CLI incompatibility." }, { - "left": "issue:43122", - "right": "issue:46209", + "left": "issue:44327", + "right": "issue:47059", "accept": false, - "reason": "Tokenizer behavior across versions and FP8 inference on B200 are entirely different issues." + "reason": "QA span decoding kth-out-of-bounds and CLI _add_completion missing are unrelated." }, { "left": "issue:46614", "right": "issue:46932", "accept": false, - "reason": "Stale phonemizer backend state and Gemma4 quantized embedder dtype casting are unrelated subsystems and failures." + "reason": "Stale phonemizer backend state and Gemma4 quantized embedder dtype casting are different subsystems." }, { - "left": "issue:44327", - "right": "issue:47059", + "left": "issue:43122", + "right": "issue:46209", "accept": false, - "reason": "A QA top-k crash and a CLI attribute error from hub version skew do not share the same code path." + "reason": "Tokenizer regression across versions is unrelated to FP8 inference failure on B200." }, { "left": "issue:43475", "right": "issue:46305", "accept": false, - "reason": "SAM3 output attribute missing and Moonshine streaming mask off-by-one are different model bugs." + "reason": "Missing SAM3 vision encoder attribute and MoonshineStreaming off-by-one are different code paths." }, { - "left": "issue:43232", - "right": "issue:45834", + "left": "issue:41628", + "right": "issue:42757", "accept": false, - "reason": "Generation kwargs handling and Kosmos2.5 OCR index overflow are not the same defect." + "reason": "Two separate import errors for different symbols and dependencies; not the same underlying fix." }, { - "left": "issue:41628", - "right": "issue:42757", + "left": "issue:43232", + "right": "issue:45834", "accept": false, - "reason": "Both are import errors, but for different symbols from different dependencies; no shared underlying fix." + "reason": "Generation sync_gpus state handling and Kosmos2.5 OCR index error are unrelated." }, { "left": "issue:30990", "right": "issue:38175", "accept": false, - "reason": "Sentence Transformers loading hang and SIGLIP2 zero-probability output are unrelated model issues." + "reason": "Sentence Transformers loading hang and siglip2 zero probabilities are different model-specific issues." }, { "left": "issue:45071", "right": "issue:46082", "accept": false, - "reason": "Both involve config validation in v5, but PretrainedConfig type checking and LlamaConfig head_dim rejection are different concrete problems." - }, - { - "left": "issue:44295", - "right": "issue:44623", - "accept": false, - "reason": "Reading a buffered position_ids tensor and processor.save_pretrained missing files are separate serialization/model API issues." + "reason": "PretrainedConfig type checking break and LlamaConfig head_dim validation are distinct config bugs." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Both are model compatibility regressions, but one is meta initialization for old InternVL2 remote-code checkpoints and the other is incorrect visual encoder keys on save_pretrained." + "reason": "Remote-code meta-init incompatibility and Qwen3.5 visual-key saving regression are not the same bug." }, { - "left": "issue:46369", - "right": "issue:46909", + "left": "issue:44295", + "right": "issue:44623", "accept": false, - "reason": "Memory overflow in sam3-litetext and a broken DIA documentation example are not the same bug." + "reason": "Buffer access error for position_ids and processor.save_pretrained missing files are unrelated." }, { "left": "issue:44190", "right": "issue:46082", "accept": false, - "reason": "Local dataset loading in a training script and LlamaConfig head_dim validation are unrelated." - }, - { - "left": "issue:43232", - "right": "issue:46424", - "accept": false, - "reason": "One is a generation-kwarg bug after sync_gpus; the other is cache config max_cache_length being honored only for quantized cache." + "reason": "Local dataset loading in an example script is unrelated to LlamaConfig head_dim validation." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "FP8 save_pretrained for MoE and the default max_shard_size are not the same issue." - }, - { - "left": "issue:42907", - "right": "issue:46050", - "accept": false, - "reason": "Both involve save/load of quantized models, but dequantized Ministrals/Devstrals and sinq round-tripping are different quantization workflows." - }, - { - "left": "issue:43408", - "right": "issue:46402", - "accept": false, - "reason": "Model-type warning for sam3_tracker and missing text encoder weights in Sam3Model are separate SAM3 bugs." - }, - { - "left": "issue:44360", - "right": "issue:46762", - "accept": false, - "reason": "A missing ReLU in the DSA indexer and numerical differences in Minimax MSA attention are different code paths." + "reason": "FP8 MOE save bug and the 50GB max_shard_size question are not the same issue." }, { - "left": "issue:43986", - "right": "issue:46726", + "left": "issue:46369", + "right": "issue:46909", "accept": false, - "reason": "Both are backend-missing AutoProcessor failures, but one is torchvision missing for a video model and the other is an unrecognized image processor error; not the same concrete bug." + "reason": "SAM3 memory overflow and a broken DIA documentation example are unrelated." }, { - "left": "issue:45381", - "right": "issue:46305", + "left": "issue:45071", + "right": "issue:45092", "accept": false, - "reason": "Qwen2.5-VL video position IDs and Moonshine sliding window masking are unrelated sequence-handling bugs." + "reason": "Both are v5 migration issues, but one is type-checking and the other is remote-code checkpoint meta initialization." }, { - "left": "issue:44263", - "right": "issue:45092", + "left": "issue:44295", + "right": "issue:46424", "accept": false, - "reason": "A torch.split return-value issue in GlmMoeDsaIndexer and an InternVL2 meta-init incompatibility do not match." + "reason": "position_ids buffer read failure and cache max_cache_length semantics are unrelated." }, { - "left": "issue:44263", - "right": "issue:44373", + "left": "issue:42907", + "right": "issue:46050", "accept": false, - "reason": "DSA indexer split semantics and a docstring typo about position_ids are clearly different issues." + "reason": "Dequantized model save failure and sinq save/load breakage are both serialization-related, but for different quantization paths and symptoms." }, { - "left": "issue:44279", - "right": "issue:46328", + "left": "issue:43408", + "right": "issue:46402", "accept": false, - "reason": "A general dependency issue and a SequenceFeatureExtractor.pad performance regression are not duplicates." + "reason": "SAM3 tracker warning and SAM3 text-encoder weights not loading are different model-loading problems." } ] }, @@ -16430,15 +16334,12 @@ "issue:22355", "issue:27957", "issue:28218", - "issue:30990", "issue:39379", "issue:39401", "issue:41628", - "issue:42757", "issue:42796", "issue:42831", "issue:42886", - "issue:42994", "issue:43066", "issue:43122", "issue:43408", @@ -16454,6 +16355,7 @@ "issue:44188", "issue:44261", "issue:44263", + "issue:44279", "issue:44360", "issue:44373", "issue:44483", @@ -16461,7 +16363,6 @@ "issue:44829", "issue:44964", "issue:45092", - "issue:45278", "issue:45310", "issue:45381", "issue:45406", @@ -16470,29 +16371,38 @@ "issue:46002", "issue:46172", "issue:46305", + "issue:46328", "issue:46369", "issue:46655", + "issue:46726", + "issue:46762", "issue:46909", "issue:47059", "issue:47217" ], "soft_pairs": [ + "issue:43986|issue:46726", + "issue:44360|issue:46762", + "issue:44263|issue:45092", + "issue:45381|issue:46305", + "issue:44263|issue:44373", + "issue:44279|issue:46328", "issue:22355|issue:27957", - "issue:46305|issue:46655", "issue:41628|issue:43824", + "issue:46305|issue:46655", "issue:43784|issue:45600", "issue:22355|issue:28218", "issue:42831|issue:45310", "issue:44483|issue:45406", "issue:43576|issue:43784", - "issue:43756|issue:44060", "issue:44964|issue:47059", + "issue:43756|issue:44060", "issue:42831|issue:44038", "issue:44188|issue:44373", "issue:44261|issue:45092", "issue:44263|issue:44360", - "issue:43986|issue:47217", "issue:44829|issue:46002", + "issue:43986|issue:47217", "issue:43644|issue:46305", "issue:42796|issue:43122", "issue:42831|issue:45753", @@ -16501,207 +16411,222 @@ "issue:43408|issue:46369", "issue:45381|issue:46909", "issue:42886|issue:43066", - "issue:39379|issue:46172", - "issue:42886|issue:42994", - "issue:27957|issue:30990", - "issue:42757|issue:45278" + "issue:39379|issue:46172" ], - "prepared_review_unit_hash": "46b04855a13e31403b651f61f948703d0256e8475534b47b949a2944e57ddf56", + "prepared_review_unit_hash": "249533a50a3de758b398d23cb098a37545b262c1f7dd343012b3c00c77dd373d", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 35711, - "estimated_input_tokens": 8928, - "estimated_eval_tokens": 18112 + "soft_pair_count": 31, + "serialized_chars": 36752, + "estimated_input_tokens": 9188, + "estimated_eval_tokens": 18632 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No soft pair appears to be the same underlying bug/change; most are related only by broad subsystem or theme (tokenizers, imports, vision-language models) but differ in concrete failure mode, code path, or affected model. There are no pull requests in the set.", - "confidence": 0.93, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: most items are only loosely similar tokenizer/vision/model regression reports, with no clear same-bug duplicate pair. The closest matches still differ in concrete code path or model/component, so I would not merge any of the soft edges.", + "confidence": 0.29, + "canonical_issue_reason": "Issue 43122 is the broadest version-level tokenizer regression in the set and the best representative of the tokenizer subtheme, though the cluster overall is too mixed for a strong canonical choice.", "canonical_pr_reason": null, - "best_issue_reason": "issue:47217 is the strongest standalone report: it has a clear, reproducible VL-model failure mode, a specific trigger (image placeholder in user text), and likely broad user impact.", + "best_issue_reason": "Issue 43122 is the most general and reusable issue title among the tokenizer-related reports; it best captures a family of tokenization regressions, even if this cluster is not a clean duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:22355", - "right": "issue:27957", + "left": "issue:43986", + "right": "issue:46726", "accept": false, - "reason": "Different problems: missing transformers.onnx module vs XLM-Roberta Flash Attention 2 behavior." + "reason": "Both are AutoProcessor/vision-backend failures, but they describe different missing-backend behaviors and different error paths." }, { - "left": "issue:46305", - "right": "issue:46655", + "left": "issue:44360", + "right": "issue:46762", + "accept": false, + "reason": "Different bugs: one is a missing ReLU in the DSA indexer, the other is numerical divergence in MSA attention." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "Unrelated code paths: torch.split behavior in GlmMoeDsaIndexer vs remote-code/meta-init incompatibility in InternVL2." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Both involve video processing, but one is wrong vision_position_ids and the other is a sliding-window mask off-by-one." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "A runtime tensor-splitting bug and a docstring issue are not the same underlying bug." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "Generic dependency issue vs a padding performance inefficiency; not the same defect." + }, + { + "left": "issue:22355", + "right": "issue:27957", "accept": false, - "reason": "Both are model/video bugs, but the concrete logic differs: sliding-window off-by-one vs padding calculation." + "reason": "Different import/model issues: missing transformers.onnx module vs XLMRoberta Flash Attention 2 support." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but for unrelated symbols and models; no shared code path is evident." + "reason": "Both are import errors, but for different symbols and likely different missing exports/code paths." + }, + { + "left": "issue:46305", + "right": "issue:46655", + "accept": false, + "reason": "Both are video-related, but one is a mask off-by-one and the other is a padding-calculation error." }, { "left": "issue:43784", "right": "issue:45600", "accept": false, - "reason": "Sentence-transformers import NameError and removed Sam3LiteText config references are unrelated regressions." + "reason": "Different failures: missing nn import in sentence-transformers integration vs stale auto_mappings references to removed configs." }, { "left": "issue:22355", "right": "issue:28218", "accept": false, - "reason": "Module import failure and tokenizer spacing bug are not the same issue." + "reason": "Different tokenizer/module problems; one is a missing module import, the other is an extra-space tokenization bug." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FineGrainedFP8 accuracy regression is unrelated to Qwen3.5 MoE from_pretrained failure." + "reason": "Both mention model/precision regressions, but the concrete failures and affected components differ." }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "Chat/completions API rejection and Gemma4Processor missing _tokenizer are different failure surfaces." + "reason": "Chat/completions request handling is unrelated to Gemma4Processor's missing _tokenizer crash." }, { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "Transformers env CLI breakage and sentence-transformers import NameError are distinct." + "reason": "CLI env command breakage and an import-time NameError are distinct failures." }, { - "left": "issue:43756", - "right": "issue:44060", + "left": "issue:44964", + "right": "issue:47059", "accept": false, - "reason": "RoPE-layer mismatch and tied-weights warning are different model-construction bugs." + "reason": "Model loading failure and CLI test failure with huggingface_hub are different issues." }, { - "left": "issue:44964", - "right": "issue:47059", + "left": "issue:43756", + "right": "issue:44060", "accept": false, - "reason": "Phi-4 multimodal loading error is unrelated to CLI completion breakage from huggingface_hub." + "reason": "Smollm3 RoPE-layer mismatch and Qwen3-Next tied-weights warning are separate model-specific bugs." }, { "left": "issue:42831", "right": "issue:44038", "accept": false, - "reason": "FP8 accuracy issue and Qwen3-VL-Moe bug are different code paths and symptoms." + "reason": "FineGrainedFP8 accuracy regression is not the same as the Qwen3-VL-Moe bug." }, { "left": "issue:44188", "right": "issue:44373", "accept": false, - "reason": "Attention-kernel divergence under torch.compile is unrelated to a wrong docstring." + "reason": "Attention-kernel divergence and a wrong docstring are not the same bug." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "Missing rms_norm_eps in MLA layernorm and remote-code meta-init incompatibility are not the same bug." + "reason": "Different problems: missing rms_norm_eps in MLA q_a_layernorm vs meta-init incompatibility for old InternVL2 checkpoints." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Both mention DSA/GLM-MoE internals, but the actual defects differ: split semantics vs missing ReLU." + "reason": "Same model family, but different functions and symptoms; not one underlying defect." }, { - "left": "issue:43986", - "right": "issue:47217", + "left": "issue:44829", + "right": "issue:46002", "accept": false, - "reason": "AutoProcessor crash without torchvision and VL image-placeholder handling are separate video/vision issues." + "reason": "Training instability under flash_attention_3 and NaNs after checkout/init are not the same concrete bug." }, { - "left": "issue:44829", - "right": "issue:46002", + "left": "issue:43986", + "right": "issue:47217", "accept": false, - "reason": "Flash Attention 3 degenerate training and NaNs after checkout/init are different training-stability problems." + "reason": "Both are VL/processor issues, but one is about missing torchvision during loading and the other about image placeholders in user text." }, { "left": "issue:43644", "right": "issue:46305", "accept": false, - "reason": "Non-persistent buffer junk and sliding-window off-by-one are unrelated implementation bugs." + "reason": "Non-persistent buffer initialization junk and a video mask off-by-one are unrelated." }, { "left": "issue:42796", "right": "issue:43122", "accept": false, - "reason": "Model-specific BPE marker handling and general version-to-version tokenization change are not clearly the same root cause." + "reason": "Tokenizer BPE-marker handling and broader v4-to-v5 tokenization differences are related only at a high level, not the same bug." }, { "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "FP8 accuracy regression and Qwen3_5 infinite-loop-on-image are different model behaviors." + "reason": "Precision/accuracy issue and an infinite-loop image-triggered bug are different failure modes." }, { "left": "issue:43540", "right": "issue:46909", "accept": false, - "reason": "Qwen3OmniMoe video processing ValueError is unrelated to a broken DIA documentation example." + "reason": "Qwen3OmniMoe video-input processing failure is unrelated to a broken documentation example." }, { "left": "issue:39401", "right": "issue:44568", "accept": false, - "reason": "Qwen3 offset_mapping and mdeberta-v3 add_special_tokens BOS/EOS regression are distinct tokenizer bugs." + "reason": "Wrong offset_mapping and missing BOS/EOS insertion are different tokenizer defects." }, { "left": "issue:43408", "right": "issue:46369", "accept": false, - "reason": "SAM3 tracker/model-type warning and sam3-litetext memory overflow are different issues." + "reason": "Model-type warning on instantiation and memory overflow in sam3-litetext are not the same issue." }, { "left": "issue:45381", "right": "issue:46909", "accept": false, - "reason": "Qwen2.5-VL video vision_position_ids bug is unrelated to DIA docs." + "reason": "A multimodal video-position bug and a docs example bug do not share the same underlying fix." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Offline cache loading failure and wrong tokenizer decoder type are separate tokenizer/runtime regressions." + "reason": "Offline cache loading and wrong tokenizer decoder type are separate tokenizer-related regressions." }, { "left": "issue:39379", "right": "issue:46172", "accept": false, - "reason": "MLLA mask handling and DeepSeek-V4 tensor transpose/casting are different low-level bugs." - }, - { - "left": "issue:42886", - "right": "issue:42994", - "accept": false, - "reason": "Tokenizer cache/offline loading and quantized model saving are unrelated failures." - }, - { - "left": "issue:27957", - "right": "issue:30990", - "accept": false, - "reason": "Flash Attention 2 on XLMRoberta and Sentence Transformers loading hang are distinct problems." - }, - { - "left": "issue:42757", - "right": "issue:45278", - "accept": false, - "reason": "A single missing huggingface_hub symbol is not the same as broad import errors after a major upgrade." + "reason": "Different tensor/masking bugs in different models; too far apart to be duplicates." } ] }, @@ -16726,17 +16651,19 @@ "cluster_id": "cluster-10000-540", "nodes": [ "issue:22355", - "issue:29127", + "issue:27957", "issue:29159", "issue:30990", "issue:33357", "issue:39401", "issue:41628", "issue:42371", + "issue:42757", + "issue:42886", "issue:42907", + "issue:42994", "issue:43116", "issue:43122", - "issue:43278", "issue:43329", "issue:43408", "issue:43452", @@ -16753,8 +16680,6 @@ "issue:44295", "issue:44297", "issue:44351", - "issue:44360", - "issue:44485", "issue:44488", "issue:44521", "issue:44625", @@ -16774,14 +16699,17 @@ "issue:47059" ], "soft_pairs": [ + "issue:42886|issue:42994", + "issue:27957|issue:30990", + "issue:42757|issue:45278", "issue:43646|issue:46002", "issue:30990|issue:33357", "issue:43329|issue:47059", "issue:43122|issue:46326", - "issue:44295|issue:44297", "issue:44625|issue:44829", - "issue:42907|issue:44964", + "issue:44295|issue:44297", "issue:22355|issue:29159", + "issue:42907|issue:44964", "issue:44117|issue:44704", "issue:39401|issue:44521", "issue:41628|issue:45278", @@ -16796,236 +16724,227 @@ "issue:43824|issue:44351", "issue:42907|issue:43937", "issue:44038|issue:44297", - "issue:44038|issue:44295", "issue:43116|issue:45200", "issue:45216|issue:45478", + "issue:44038|issue:44295", "issue:44038|issue:44222", - "issue:43122|issue:43906", "issue:44995|issue:46697", - "issue:44360|issue:44485", - "issue:43278|issue:44945", - "issue:22355|issue:29127" + "issue:43122|issue:43906" ], - "prepared_review_unit_hash": "92848e44085a33c2c2a8aa1c8281331cd66a118c183a1c3913f9d5c1e4b00391", + "prepared_review_unit_hash": "3cd174b84293227eed160d6229b2b0da048469d4cc85d71a7b1d0d23dd627e0e", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 31, - "serialized_chars": 36125, - "estimated_input_tokens": 9032, - "estimated_eval_tokens": 18320 + "serialized_chars": 36036, + "estimated_input_tokens": 9009, + "estimated_eval_tokens": 18274 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most items are unrelated issues spanning different subsystems (tokenization, multimodal, import/export, CLI, model loading, and platform-specific bugs). I only see one plausible duplicate pair around Qwen3.5 save/load regression; the rest are not the same underlying problem.", - "confidence": 0.68, - "canonical_issue_reason": null, + "summary": "Most candidates are clearly unrelated issues spanning different models, features, and failure modes. The only plausible duplicate-like pair is the Qwen3.5 save/load regression around transformers 5.4.0, which appears to describe the same underlying serialization bug from two directions.", + "confidence": 0.89, + "canonical_issue_reason": "issue:45216 is the strongest representative issue here: it describes a concrete, version-bounded Qwen3.5 regression in save/load behavior and is the clearest root-cause-style report among the only plausible duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:45216 is the best standalone issue candidate because it is specific, reproducible, and tied to a clear regression boundary (since 5.4.0).", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Both involve saving/loading, but one is tokenizer cache loading offline and the other is quantized model saving; different code paths and failure modes." + }, + { + "left": "issue:27957", + "right": "issue:30990", + "accept": false, + "reason": "Different problems: Flash Attention 2 support for XLM-Roberta vs Sentence Transformers getting stuck while loading." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "The second is a broad post-upgrade import-error report, not the same concrete missing symbol bug as is_offline_mode." + }, { "left": "issue:43646", "right": "issue:46002", "accept": false, - "reason": "Both are model initialization failures, but one is about custom model init breaking in 5.0 and the other about NaNs in classification heads after checkout/init; different symptoms and likely different code paths." + "reason": "Custom model initialization breakage and NaNs in classification heads are distinct bugs with different symptoms." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "Loading stuck for Sentence Transformers and a MacOS bus error with a CLIP community model are unrelated failure modes." + "reason": "Loading hang for Sentence Transformers and a MacOS bus error on a CLIP model are unrelated failures." }, { "left": "issue:43329", "right": "issue:47059", "accept": false, - "reason": "One is a multimodal token-counting bug in video processing; the other is a CLI compatibility error with huggingface_hub/typer. No shared code path." + "reason": "Multimodal token counting bug vs CLI incompatibility with huggingface_hub; different subsystems entirely." }, { "left": "issue:43122", "right": "issue:46326", "accept": false, - "reason": "Different tokenizer bugs: version-to-version tokenization drift versus mistral_common continue_final_message behavior." - }, - { - "left": "issue:44295", - "right": "issue:44297", - "accept": false, - "reason": "Reading a registered buffer error and tokenizer_class mismatch on save_pretrained are distinct serialization issues." + "reason": "Both are tokenizer-related, but one is an offset/differences regression and the other is a mistral_common chat-template bug." }, { "left": "issue:44625", "right": "issue:44829", "accept": false, - "reason": "Qwen3.5 num_labels propagation and flash_attention_3 degenerate training are unrelated." + "reason": "Qwen3.5 num_labels propagation and flash_attention_3 training degeneration are unrelated issues." }, { - "left": "issue:42907", - "right": "issue:44964", + "left": "issue:44295", + "right": "issue:44297", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals and loading Phi-4 multimodal are different model-specific bugs." + "reason": "One is a position_ids buffer/read problem, the other is tokenizer_class metadata written by save_pretrained." }, { "left": "issue:22355", "right": "issue:29159", "accept": false, - "reason": "ONNX import error and slow/fast tokenizer inconsistency are unrelated." + "reason": "Missing transformers.onnx import and slow/fast tokenizer inconsistency are different topics." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Quantized/dequantized model saving and loading a multimodal Phi model are unrelated." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES None handling and AutoProcessor cached_file kwargs are different loader bugs." + "reason": "TOKENIZER_MAPPING_NAMES returning None and AutoProcessor kwargs not reaching cached_file affect different layers of the loading stack." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Qwen3 offset_mapping and multimodal assistant_masks are different tokenizer/template behaviors." + "reason": "Wrong offset_mapping and zero assistant_masks in multimodal chat templates are different tokenizer behaviors." }, { "left": "issue:41628", "right": "issue:45278", "accept": false, - "reason": "An AutoImageProcessor import error may be one example of broader import failures, but the issues are not clearly the same concrete bug." + "reason": "The latter is a broad upgrade regression, not the same specific AutoImageProcessor import failure." }, { "left": "issue:39401", "right": "issue:44488", "accept": false, - "reason": "Different tokenization offset bug versus a model loading failure for cjvt/sleng-bert." + "reason": "Offset mapping bugs and failure to load a specific tokenizer are unrelated." }, { "left": "issue:42907", "right": "issue:44945", "accept": false, - "reason": "Model save/dequantization issue and pipeline-parallelism output corruption are unrelated." + "reason": "Quantized model save failure and incorrect LLM output under pipeline parallelism do not share the same code path." }, { "left": "issue:43976", "right": "issue:44945", "accept": false, - "reason": "Python version compatibility regression and pipeline parallelism output issue are different problems." + "reason": "Python version compatibility and pipeline-parallel output corruption are different bugs." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Different tokenizer/template bugs: tokenization drift versus multimodal assistant mask generation." + "reason": "General tokenization differences versus multimodal assistant mask generation are not the same underlying bug." }, { "left": "issue:44704", "right": "issue:45290", "accept": false, - "reason": "Passing kwargs to cached_file is unrelated to chat template crashes with tool-call assistant messages." + "reason": "AutoProcessor cached_file forwarding and apply_chat_template crashing on tool-call messages are distinct issues." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Saving dequantized models and gguf_file loading failures affect different code paths." + "reason": "Dequantized model saving and gguf_file loading regressions are separate features and failure modes." }, { "left": "issue:43408", "right": "issue:43475", "accept": false, - "reason": "Both involve SAM 3 Video, but one is a model-type warning and the other is a missing attribute during vision encoding; not the same bug." + "reason": "Both mention SAM 3 video, but one is a model-type warning and the other is a missing attribute during vision encoder output handling; not the same bug." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 API warning and GlmMoeDsaIndexer torch.split behavior are unrelated." + "reason": "TF32 settings warning and GlmMoeDsa torch.split behavior are unrelated." }, { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Different missing imports for different symbols; no shared underlying issue is evident." + "reason": "Different missing imports: Qwen2_5_VLForConditionalGeneration vs HybridCache." }, { "left": "issue:42907", "right": "issue:43937", "accept": false, - "reason": "Model save/load issue and invalid GenerationConfig for GLM-5 are unrelated." + "reason": "Quantized save failure and invalid GenerationConfig for GLM-5 are unrelated." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "Qwen3-VL-Moe load bug and tokenizer save_pretrained class mismatch are not the same problem." - }, - { - "left": "issue:44038", - "right": "issue:44295", - "accept": false, - "reason": "Qwen3-VL-Moe loading and position_ids buffer access are separate failures." + "reason": "Qwen3-VL-Moe compatibility and tokenizer_class metadata mismatch are separate problems." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Example script multi-label output bug and Gemma 4 mm_token_type_ids defaulting are different issues." + "reason": "Example-script multi-label classification behavior and Gemma 4 mm_token_type_ids defaults are different concerns." }, { "left": "issue:45216", "right": "issue:45478", "accept": true, - "reason": "Both are Qwen3.5 MoE regressions tied to transformers 5.4.0+ and appear to concern the same save/load serialization path, just with different user-facing symptoms." + "reason": "Both are Qwen3.5 regressions tied to transformers 5.4.0+ save/load behavior; one reports an incorrect saved checkpoint and the other a from_pretrained failure, which plausibly share the same underlying serialization/config bug." }, { "left": "issue:44038", - "right": "issue:44222", - "accept": false, - "reason": "Qwen3-VL-Moe loading and FP8 save_pretrained for moe are different model-specific bugs." - }, - { - "left": "issue:43122", - "right": "issue:43906", + "right": "issue:44295", "accept": false, - "reason": "Tokenizer version drift and an isolated reproduction of another issue are not the same concrete bug." + "reason": "Qwen3-VL-Moe version breakage and a position_ids buffer access issue are unrelated." }, { "left": "issue:44995", "right": "issue:46697", "accept": false, - "reason": "GlmMoeDsa second forward cache bug and Swin2SR image over-padding are unrelated." - }, - { - "left": "issue:44360", - "right": "issue:44485", - "accept": false, - "reason": "DSA indexer ReLU discussion and GLM-5 RoPE implementation are separate model-internal concerns." + "reason": "GlmMoeDsa stale cache on second forward pass and Swin2SR over-padding are different model-specific bugs." }, { - "left": "issue:43278", - "right": "issue:44945", - "accept": false, - "reason": "Embedding dtype mismatch during evaluate and incorrect LLM output under pipeline parallelism are different bugs." - }, - { - "left": "issue:22355", - "right": "issue:29127", + "left": "issue:43122", + "right": "issue:43906", "accept": false, - "reason": "Transformers.onnx import error and layoutlmv3 box info error are unrelated." + "reason": "A tokenization regression and a reproduction of an unrelated external issue are not the same bug." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays appropriately conservative. The one accepted Qwen3.5 pair is plausibly related by model family, version window, and save/load path, while the rest of the soft-edge verdicts avoid overclaiming duplicates and mostly explain why the symptoms look different." + "feedback": "The summary is grounded in the packet: issue:45216 is indeed the clearest version-bounded Qwen3.5 save/load regression candidate by title, and the only accepted soft-edge pair (45216/45478) is a conservative same-area match rather than an overreach. The remaining verdicts are appropriately cautious and the reasons stay within the evidence provided by the titles." }, "evaluator_used": true, "retried": false, @@ -17046,7 +16965,9 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ + "issue:22355", "issue:28348", + "issue:29127", "issue:36296", "issue:38617", "issue:39379", @@ -17055,30 +16976,28 @@ "issue:42222", "issue:42673", "issue:42757", - "issue:42796", "issue:42898", "issue:43066", "issue:43116", "issue:43232", + "issue:43278", "issue:43493", "issue:43611", - "issue:43937", "issue:43976", - "issue:44075", "issue:44188", "issue:44351", + "issue:44360", "issue:44483", "issue:44485", "issue:44610", "issue:44625", "issue:44779", - "issue:44936", "issue:44938", + "issue:44945", "issue:45072", "issue:45083", "issue:45278", "issue:45357", - "issue:45362", "issue:45446", "issue:45478", "issue:45753", @@ -17088,7 +17007,6 @@ "issue:46328", "issue:46614", "issue:46640", - "issue:46710", "issue:46762", "issue:46877", "issue:46909", @@ -17096,241 +17014,224 @@ "issue:47312" ], "soft_pairs": [ + "issue:44360|issue:44485", + "issue:43278|issue:44945", + "issue:22355|issue:29127", "issue:46172|issue:46762", "issue:42175|issue:44938", "issue:28348|issue:46640", "issue:42222|issue:47312", - "issue:39401|issue:46614", "issue:39379|issue:43232", "issue:43493|issue:45753", - "issue:44188|issue:46328", + "issue:39401|issue:46614", "issue:42757|issue:44351", + "issue:44188|issue:46328", "issue:42673|issue:45072", "issue:36296|issue:43116", "issue:45357|issue:45478", "issue:46231|issue:46877", - "issue:42898|issue:46614", "issue:44483|issue:47217", + "issue:42898|issue:46614", "issue:43976|issue:45446", - "issue:44610|issue:46909", "issue:44625|issue:46002", + "issue:44610|issue:46909", "issue:45083|issue:45446", "issue:43611|issue:46002", "issue:38617|issue:45278", "issue:43116|issue:43493", - "issue:43066|issue:44779", - "issue:44075|issue:44936", - "issue:43937|issue:44485", - "issue:45362|issue:45478", - "issue:42673|issue:42796", - "issue:43066|issue:46710" + "issue:43066|issue:44779" ], - "prepared_review_unit_hash": "9b03d8670939580ecb0b494c71df92c4c4cebec7afbabf3f90ccdcd0cc3cf2b3", + "prepared_review_unit_hash": "0f15dc0c3e0d87950a575ef5ed0f2a01babc582ba4215787f7ae839357985066", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 35499, - "estimated_input_tokens": 8875, - "estimated_eval_tokens": 18006 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 34379, + "estimated_input_tokens": 8595, + "estimated_eval_tokens": 17446 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous and does not look like one duplicate set. The only strong duplicate-like link is the tokenizer v5 regression pair (43066 \u2194 44779); the rest are distinct bugs/features across unrelated subsystems.", - "confidence": 0.84, - "canonical_issue_reason": "issue:43066 is the best anchor because it describes the broad v5 tokenizer decoder regression that plausibly explains the model-specific symptom in 44779.", + "summary": "The set is a heterogeneous mix of unrelated bugs, regressions, and docs/import issues. None of the soft pairs look like the same underlying change or bug, so I reject all soft edges. The broadest issue is the multi-import-regression report, which is the best umbrella representative here.", + "confidence": 0.92, + "canonical_issue_reason": "issue:45278 is the broadest, most umbrella-style report in the set ('many import errors after update'), making it the best canonical issue candidate among otherwise unrelated reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43066 is the most reusable and central bug report among the only plausible duplicate pair, so it best represents the cluster if one issue must be chosen.", + "best_issue_reason": "issue:45278 is the most general and reusable problem statement in the cluster; it can serve as the closest thing to a representative issue even though the cluster is not a true duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "Unrelated topics: DSA indexer/ReLU discussion vs GLM-5 RoPE implementation." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Different bug classes: dtype drift between train/eval vs incorrect output under pipeline parallelism." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "Import failure for transformers.onnx is unrelated to LayoutLMv3 box-content error messaging." + }, { "left": "issue:46172", "right": "issue:46762", "accept": false, - "reason": "Different problems in different attention math paths: deepseek-v4 dtype/transpose issue vs Minimax attention numerical drift." + "reason": "Different models and failure modes: Deepseek-v4 tensor expression vs Minimax attention numerical mismatch." }, { "left": "issue:42175", "right": "issue:44938", "accept": false, - "reason": "One is a package/backend dependency installation issue; the other is a Python version loading failure." + "reason": "Both are environment/runtime problems, but one is missing TensorFlow from install extras and the other is a Python 3.14 load failure; not the same bug." }, { "left": "issue:28348", "right": "issue:46640", "accept": false, - "reason": "Both mention Flash Attention 2, but one is a GPT2 support request and the other is a separate runtime issue report." + "reason": "Both mention flash attention 2, but one is a GPT-2 support request and the other is an unspecified issue; too vague to merge as the same bug." }, { "left": "issue:42222", "right": "issue:47312", "accept": false, - "reason": "VitPose model breakage and kernel compatibility are too broad and do not point to the same concrete failure." - }, - { - "left": "issue:39401", - "right": "issue:46614", - "accept": false, - "reason": "Different tokenizers and different bugs: Qwen3 offset mapping vs Wav2Vec2 phonemizer language state." + "reason": "Vitpose model breakage is unrelated to kernel compatibility with 0.16." }, { "left": "issue:39379", "right": "issue:43232", "accept": false, - "reason": "mllama masking logic and generation kwargs after sync_gpus are unrelated code paths." + "reason": "Mllama masking logic and generation kwargs after sync_gpus are separate code paths." }, { "left": "issue:43493", "right": "issue:45753", "accept": false, - "reason": "Both are vision-model regressions, but one is a SigLIP2 implementation mismatch and the other is an infinite-loop image bug." + "reason": "Different models and symptoms: SigLIP2 JAX parity mismatch vs Qwen3_5 image-triggered infinite loop." }, { - "left": "issue:44188", - "right": "issue:46328", + "left": "issue:39401", + "right": "issue:46614", "accept": false, - "reason": "Attention-kernel divergence and SequenceFeatureExtractor padding performance are unrelated." + "reason": "Tokenizer offset mapping bug is unrelated to stale espeak backend language state in Wav2Vec2PhonemeCTCTokenizer." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Separate import errors from different modules and different missing symbols." + "reason": "Both are import errors, but for different missing symbols from different packages; not the same concrete failure." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "Attention-kernel divergence under torch.compile is unrelated to a padding performance issue." }, { "left": "issue:42673", "right": "issue:45072", "accept": false, - "reason": "VRAM leak in Qwen3 CausalLM threads is unrelated to bfloat16 dtype mismatches in other models." + "reason": "VRAM leak in Qwen3ForCausalLM is unrelated to dtype mismatches in inference tests." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Tensor parallel training bug and empty multi-label classification results are different failures." + "reason": "Tensor-parallel training bug and empty multi-label results in an example script are different behaviors with no concrete shared code path." }, { "left": "issue:45357", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5, but one is save_pretrained key serialization and the other is a from_pretrained error." + "reason": "Both involve Qwen3.5, but one is save_pretrained key serialization and the other is from_pretrained loading failure; too different to be duplicates." }, { "left": "issue:46231", "right": "issue:46877", "accept": false, - "reason": "A docs parameter-name mismatch is not the same as auto-docstring printing diagnostics at import time." + "reason": "Documentation parameter name mismatch is unrelated to auto_docstring printing diagnostics at import time." }, { - "left": "issue:42898", - "right": "issue:46614", + "left": "issue:44483", + "right": "issue:47217", "accept": false, - "reason": "Tokenization cleanup behavior changes and phonemizer backend state are unrelated." + "reason": "Chat completions request rejection is not the same as VL models failing on image placeholders in user text." }, { - "left": "issue:44483", - "right": "issue:47217", + "left": "issue:42898", + "right": "issue:46614", "accept": false, - "reason": "Chat-completions request handling and VL image-placeholder handling are different API paths." + "reason": "Tokenization cleanup behavior in v5 is unrelated to stale phonemizer backend language switching." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "Both are version-check issues, but they affect different dependencies and code paths." + "reason": "Both are version-compatibility checks, but they concern different runtimes and different import sites." }, { - "left": "issue:44610", - "right": "issue:46909", + "left": "issue:44625", + "right": "issue:46002", "accept": false, - "reason": "Processor image-size mismatch and documentation example breakage are unrelated." + "reason": "Configuration propagation bug in Qwen3.5 is unrelated to NaNs in classification heads after checkout/init." }, { - "left": "issue:44625", - "right": "issue:46002", + "left": "issue:44610", + "right": "issue:46909", "accept": false, - "reason": "num_labels propagation in Qwen3.5 config is distinct from NaNs in classification heads." + "reason": "Processor input-size mismatch is unrelated to a broken documentation example." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "A helper-length calculation bug and a PyTorch version check bug are not the same issue." + "reason": "Helper output-length logic in qwen3_omni_moe is unrelated to a PyTorch version gate for flex_attention." }, { "left": "issue:43611", "right": "issue:46002", "accept": false, - "reason": "Model loading with base_model_prefix and classification-head NaNs are different regressions." + "reason": "Base-model-prefix loading regression and NaNs in classification heads are not the same failure." }, { "left": "issue:38617", "right": "issue:45278", "accept": false, - "reason": "A specific missing import may be one symptom of broader import breakage, but this is not enough to treat them as the same bug." + "reason": "The latter is a broad umbrella of many import regressions; the former is one specific missing symbol and not clearly the same bug." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Example-script multi-label output and SigLIP2 implementation differences are unrelated." + "reason": "Example-script classification output issue is unrelated to SigLIP2 implementation parity." }, { "left": "issue:43066", "right": "issue:44779", - "accept": true, - "reason": "44779 looks like a model-specific manifestation of the same v5 tokenizer decoder regression described generically in 43066." - }, - { - "left": "issue:44075", - "right": "issue:44936", - "accept": false, - "reason": "Unused SGD args and trainer.evaluate() failing after train() are different training bugs." - }, - { - "left": "issue:43937", - "right": "issue:44485", - "accept": false, - "reason": "GenerationConfig validation and GLM-5 RoPE implementation are different GLM-5 issues, not the same defect." - }, - { - "left": "issue:45362", - "right": "issue:45478", - "accept": false, - "reason": "Both mention Qwen3.5 crashes, but one is chat runtime and the other is model loading; no shared concrete bug is evident." - }, - { - "left": "issue:42673", - "right": "issue:42796", - "accept": false, - "reason": "VRAM leakage in Qwen3 CausalLM is unrelated to tokenizer BPE-marker handling." - }, - { - "left": "issue:43066", - "right": "issue:46710", "accept": false, - "reason": "The DeepSeek accuracy regression may be related to v5 behavior, but it is not the same concrete tokenizer-decoder bug as 43066." + "reason": "Both involve tokenizers and v5 regressions, but one is decoder-type selection and the other is Deepseek output correctness; not enough evidence they are the same bug." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The lone accepted pair (43066 \u2194 44779) is supported by the shared v5 tokenizer-regression theme, while the other soft pairs are reasonably rejected as distinct subsystems or failure modes based on the titles. " - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -17353,40 +17254,41 @@ "issue:42222", "issue:42371", "issue:42617", + "issue:42673", + "issue:42796", "issue:43064", + "issue:43066", "issue:43116", "issue:43208", - "issue:43408", - "issue:43441", "issue:43618", "issue:43673", "issue:43688", - "issue:43742", "issue:43824", "issue:43825", "issue:43827", "issue:43874", "issue:43937", "issue:43992", + "issue:44075", "issue:44117", "issue:44291", "issue:44355", - "issue:44380", - "issue:44466", + "issue:44485", "issue:44704", - "issue:44977", + "issue:44936", "issue:45071", "issue:45081", "issue:45092", "issue:45341", + "issue:45362", "issue:45405", "issue:45446", + "issue:45478", "issue:45563", "issue:45588", - "issue:46002", - "issue:46077", "issue:46231", "issue:46328", + "issue:46710", "issue:46726", "issue:46762", "issue:46829", @@ -17396,12 +17298,17 @@ "issue:47116" ], "soft_pairs": [ + "issue:44075|issue:44936", + "issue:45362|issue:45478", + "issue:43937|issue:44485", + "issue:42673|issue:42796", + "issue:43066|issue:46710", "issue:46726|issue:46981", "issue:44355|issue:46231", "issue:43618|issue:46829", "issue:43116|issue:43825", - "issue:43827|issue:45446", "issue:42371|issue:43874", + "issue:43827|issue:45446", "issue:43992|issue:44704", "issue:45405|issue:45446", "issue:44117|issue:45588", @@ -17410,210 +17317,205 @@ "issue:39379|issue:43688", "issue:46328|issue:47116", "issue:44117|issue:45081", - "issue:43673|issue:46328", "issue:33453|issue:39401", + "issue:43673|issue:46328", "issue:43208|issue:46894", "issue:43064|issue:43673", "issue:43937|issue:45563", "issue:42617|issue:45341", - "issue:45092|issue:46762", "issue:42222|issue:47000", - "issue:44466|issue:46002", - "issue:44380|issue:46077", - "issue:43937|issue:44977", - "issue:42222|issue:43441", - "issue:43408|issue:43742" + "issue:45092|issue:46762" ], - "prepared_review_unit_hash": "6bd06b91d2a6394640ac8accaf06a360aa0630358ae315545eb6b5a2ba6d13c1", + "prepared_review_unit_hash": "fb6288bf0bcd4d785168287fa355dfc337abaccea3a49ff5955774daa22e0c89", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35236, - "estimated_input_tokens": 8809, - "estimated_eval_tokens": 17874 + "serialized_chars": 35627, + "estimated_input_tokens": 8907, + "estimated_eval_tokens": 18070 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a true duplicate set; it mixes many unrelated regressions across tokenizers, generation, loading, docs, and model-specific behavior. The soft-similarity pairs are mostly broad thematic matches, not the same concrete bug or change.", - "confidence": 0.18, - "canonical_issue_reason": "Issue #44117 is the closest broad representative because it concerns a core from_pretrained/tokenizer-loading assumption that sits near the center of several loading-related regressions, but the overall cluster is too heterogeneous for a strong canonical duplicate.", + "summary": "This cluster is highly heterogeneous: most pairs are only loosely related by broad areas like tokenizers, v5 regressions, or model loading, but they do not describe the same underlying bug or change.", + "confidence": 0.27, + "canonical_issue_reason": "No single true canonical duplicate stands out; issue:44117 is only the best broad representative because it concerns a generic from_pretrained/tokenizer-loading assumption, while most others are model-specific or unrelated.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #44117 is the best single issue to anchor the cluster if one must be chosen, since it is a general loading-path bug rather than a model-specific or docs-only complaint.", + "best_issue_reason": "issue:44117 is the most reusable cluster anchor: it targets a core tokenizer-loading path and is broader than the other mostly narrow, model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "Different failures in different training/eval paths: unused SGD args vs evaluate() breaking after train()." + }, + { + "left": "issue:45362", + "right": "issue:45478", + "accept": false, + "reason": "Both mention Qwen3.5, but one is a chat crash and the other is a from_pretrained MoE loading error; not the same bug." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "One is invalid GenerationConfig handling, the other is RoPE implementation discussion; separate GLM-5 issues." + }, + { + "left": "issue:42673", + "right": "issue:42796", + "accept": false, + "reason": "VRAM leak in multi-threaded dataloading vs tokenizer BPE marker handling; unrelated subsystems." + }, + { + "left": "issue:43066", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer decoder type regression and an accuracy regression in a specific model are not the same code-path problem." + }, { "left": "issue:46726", "right": "issue:46981", "accept": false, - "reason": "Both are AutoProcessor/loading-related, but one is about missing vision backend handling and the other is about repeated 429 retries; different failure modes and code paths." + "reason": "AutoProcessor missing-vision-backend error and repeated HTTP 429 retries are different loading/network behaviors." }, { "left": "issue:44355", "right": "issue:46231", "accept": false, - "reason": "A runtime error on compiled Python files is unrelated to a documentation parameter-name mismatch." + "reason": "Compiled Python file errors and a set_training() docs parameter mismatch are unrelated." }, { "left": "issue:43618", "right": "issue:46829", "accept": false, - "reason": "Both mention output fields, but they affect different model heads and different incorrect assignments; not the same bug." + "reason": "Both concern output fields, but they affect different model heads and different output objects; not one duplicate bug." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "Multi-label classification returning empty results is unrelated to pipeline() wording/error-message behavior." - }, - { - "left": "issue:43827", - "right": "issue:45446", - "accept": false, - "reason": "Docs still referencing pipeline() and a PyTorch version check bug in flex_attention are unrelated issues." + "reason": "Example script multi-label empty results and a pipeline() error message about translation support are distinct issues." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 settings behavior and a missing image-processor method in GLM46V are distinct model/runtime bugs." + "reason": "TF32 API-setting guidance and a missing image-patch method in GLM46V are unrelated." }, { - "left": "issue:43992", + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs still referencing pipeline() and a PyTorch version-check bug in flex_attention are different kinds of defects." + }, + { + "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "UMT5Encoder weight-loading and AutoProcessor kwarg forwarding are different loading bugs in different layers." + "reason": "Both involve from_pretrained, but one misses model weights and the other fails to forward kwargs to cached_file." }, { "left": "issue:45405", "right": "issue:45446", "accept": false, - "reason": "A dependency version bump in docs/config is not the same as a runtime version-check bug." + "reason": "A dependency version bump to an unreleased PEFT version is unrelated to a flex_attention import/version check." }, { "left": "issue:44117", "right": "issue:45588", "accept": false, - "reason": "Tokenizer mapping returning None and a flash_attention crash on s_aux=None are unrelated code paths." + "reason": "Tokenizer mapping None handling and a flash_attention crash on s_aux=None are separate bugs in different paths." }, { "left": "issue:43824", "right": "issue:46981", "accept": false, - "reason": "An import error for a missing model class is unrelated to HTTP 429 retry behavior." + "reason": "Importing a missing model class and retry behavior on 429 responses are unrelated." }, { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "Meta-initialization arg handling and PretrainedConfig type checking are both v5-adjacent but not the same concrete defect." + "reason": "Both are v5 loading-related, but one is init_empty_weights/_is_hf_initialized and the other is PretrainedConfig type checking; not the same fix." }, { "left": "issue:39379", "right": "issue:43688", "accept": false, - "reason": "mllama mask handling and auxiliary-loss normalization are different numerical/logic bugs." + "reason": "mllama mask handling and MoE auxiliary-loss normalization are different model-specific math bugs." }, { "left": "issue:46328", "right": "issue:47116", "accept": false, - "reason": "A padding performance issue is not the same as a streaming pipeline regression." + "reason": "A SequenceFeatureExtractor.pad performance issue is unrelated to a streaming pipeline regression." }, { "left": "issue:44117", "right": "issue:45081", "accept": false, - "reason": "Tokenizer mapping None and Mistral regex patching crash involve different tokenizer-loading failures." + "reason": "Tokenizer mapping fallback and Mistral regex patch crashes involve different tokenizer code paths." }, { - "left": "issue:43673", - "right": "issue:46328", + "left": "issue:33453", + "right": "issue:39401", "accept": false, - "reason": "Missing GenerationMixin cache in chunked prefill is unrelated to SequenceFeatureExtractor padding overhead." + "reason": "Both are tokenizer-related, but one is loading regression and the other is offset_mapping for Qwen3; not the same underlying defect." }, { - "left": "issue:33453", - "right": "issue:39401", + "left": "issue:43673", + "right": "issue:46328", "accept": false, - "reason": "Both are tokenizer-related, but one is a loading regression and the other is a wrong offset_mapping issue." + "reason": "Generation cache handling and feature extractor padding performance are unrelated." }, { "left": "issue:43208", "right": "issue:46894", "accept": false, - "reason": "xLSTM training blockers and CohereASR double-shift loss bug are separate model/training issues." + "reason": "xLSTM training blockers and a CohereASR double-shift loss bug are not the same issue, even if both affect training." }, { "left": "issue:43064", "right": "issue:43673", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption is unrelated to missing cache during generation chunked_prefill." + "reason": "Optimizer-state corruption under FSDP2/PEFT is unrelated to missing GenerationMixin cache during chunked prefill." }, { "left": "issue:43937", "right": "issue:45563", "accept": false, - "reason": "Invalid GenerationConfig and a stale warning for num_return_sequences are different generation issues." + "reason": "Invalid generation config and a stale warning for num_return_sequences are separate generation issues." }, { "left": "issue:42617", "right": "issue:45341", "accept": false, - "reason": "Running 3d_parallel.py and a small testing_utils bug do not describe the same underlying problem." - }, - { - "left": "issue:45092", - "right": "issue:46762", - "accept": false, - "reason": "Meta-initialization incompatibility for remote-code checkpoints is unrelated to Minimax attention numerical differences." + "reason": "3d_parallel.py runtime failure and a small testing_utils bug do not share an underlying cause." }, { "left": "issue:42222", "right": "issue:47000", "accept": false, - "reason": "VitPose loading breakage and RTDetrLoss infeasible cost-matrix errors are unrelated model-specific failures." - }, - { - "left": "issue:44466", - "right": "issue:46002", - "accept": false, - "reason": "Serialization of tied weights and NaNs in classification heads are different bugs with different symptoms and likely fixes." - }, - { - "left": "issue:44380", - "right": "issue:46077", - "accept": false, - "reason": "GPT2 attention scaling being ignored and Gemma4 bidirectional attention masking are distinct attention-behavior bugs." - }, - { - "left": "issue:43937", - "right": "issue:44977", - "accept": false, - "reason": "Both are generation/flash-attention themed, but one is invalid config validation and the other is a model-specific flash-attention generation failure." - }, - { - "left": "issue:42222", - "right": "issue:43441", - "accept": false, - "reason": "VitPose model breakage and Ministral-3 flash-attention failure are unrelated model-specific regressions." + "reason": "VitPose model breakage and RTDetrLoss infeasible cost matrix errors are unrelated model failures." }, { - "left": "issue:43408", - "right": "issue:43742", + "left": "issue:45092", + "right": "issue:46762", "accept": false, - "reason": "A sam3 tracker/model-type warning and a MobileLLM loading KeyError are different model-loading issues." + "reason": "Meta-initialization incompatibility for old InternVL2 checkpoints and Minimax attention numerical drift are different problems." } ] }, @@ -17642,23 +17544,21 @@ "issue:39401", "issue:41628", "issue:42222", - "issue:42371", - "issue:42831", "issue:43097", "issue:43122", - "issue:43208", + "issue:43408", "issue:43421", + "issue:43441", "issue:43450", - "issue:43931", + "issue:43742", "issue:43937", "issue:43986", - "issue:44077", "issue:44220", - "issue:44261", "issue:44263", + "issue:44380", "issue:44410", + "issue:44466", "issue:44483", - "issue:44484", "issue:44493", "issue:44496", "issue:44704", @@ -17666,15 +17566,17 @@ "issue:44829", "issue:44877", "issue:44964", + "issue:44977", "issue:45071", "issue:45216", "issue:45278", "issue:45310", "issue:45341", - "issue:45440", "issue:45464", "issue:45561", "issue:45632", + "issue:46002", + "issue:46077", "issue:46231", "issue:46298", "issue:46328", @@ -17687,7 +17589,12 @@ "issue:47110" ], "soft_pairs": [ + "issue:44466|issue:46002", + "issue:44380|issue:46077", + "issue:43408|issue:43742", + "issue:43937|issue:44977", "issue:43450|issue:43986", + "issue:42222|issue:43441", "issue:39379|issue:46897", "issue:45071|issue:46512", "issue:42222|issue:46752", @@ -17699,226 +17606,207 @@ "issue:45341|issue:45561", "issue:44410|issue:45310", "issue:45071|issue:47059", - "issue:45440|issue:46328", + "issue:45341|issue:46328", "issue:39401|issue:43421", "issue:44483|issue:45464", - "issue:44704|issue:47110", - "issue:43937|issue:44496", "issue:38617|issue:41628", "issue:38617|issue:44220", + "issue:43937|issue:44496", + "issue:44704|issue:47110", "issue:43122|issue:46298", "issue:44964|issue:45278", - "issue:43097|issue:44493", - "issue:42371|issue:44261", - "issue:42371|issue:42831", - "issue:44261|issue:46762", - "issue:43208|issue:46897", - "issue:44484|issue:45310", - "issue:44077|issue:46981", - "issue:43931|issue:44410" + "issue:43097|issue:44493" ], - "prepared_review_unit_hash": "c7bca8269605790829213c50365b1e84a97f25f6f4e9e611bbf95f8060560f62", + "prepared_review_unit_hash": "e7c9e3d840222e644bab457bfe24b9a758e0bf0e0d2a172bf43e75548662f5aa", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36150, - "estimated_input_tokens": 9038, - "estimated_eval_tokens": 18332 + "soft_pair_count": 27, + "serialized_chars": 35610, + "estimated_input_tokens": 8903, + "estimated_eval_tokens": 18062 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous cluster of unrelated Transformers issues (imports, tokenization, multimodal loading, training regressions, API behavior, and docs/perf). None of the soft pairs look like the same underlying bug, so all are rejected and there is no single canonical duplicate target for the cluster.", - "confidence": 0.94, - "canonical_issue_reason": null, + "summary": "The cluster is highly heterogeneous: it mixes unrelated tokenization, model-loading, config, processor, API, and training bugs. The soft links mostly reflect broad transformer-v5 regression wording or shared subsystem names, not the same underlying defect.", + "confidence": 0.96, + "canonical_issue_reason": "No single issue is a true duplicate of the whole cluster; if forced to pick an umbrella, issue:45278 is the broadest regression report, but it still only loosely covers a subset of the import/load failures.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44704 is the most concrete, actionable bug report in the set: it names a specific code path (`AutoProcessor.from_pretrained`) and a clear failure mode (missing kwargs reaching `cached_file`).", + "best_issue_reason": "issue:45278 is the best umbrella candidate because it is the most general user-facing regression report and could absorb some related import/loading complaints, even though most items here are distinct bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44466", + "right": "issue:46002", + "accept": false, + "reason": "Different symptoms and code paths: weight serialization/tied weights vs NaNs in classification heads." + }, + { + "left": "issue:44380", + "right": "issue:46077", + "accept": false, + "reason": "Unrelated model bugs: GPT-2 attention scaling vs Gemma4 bidirectional mask construction." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "A warning about model type mismatch is not the same bug as a key error while loading a MobileLLM model." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "Different failures: invalid GenerationConfig vs flash-attention generation behavior for Qwen3.5." + }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Both involve video/multimodal processing, but one is a batched-output shape bug and the other is a torchvision-dependent loading crash; different failure modes and code paths." + "reason": "Batched video processor shape bug and missing-torchvision crash are separate problems." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Vitpose breakage and Ministral FlashAttention failure are unrelated model-specific regressions." }, { "left": "issue:39379", "right": "issue:46897", "accept": false, - "reason": "mllama mask handling and Florence2 training-loss shift are unrelated model-specific bugs." + "reason": "mllama mask handling and Florence2 training-loss shifting are different bugs, only loosely pattern-matched." }, { "left": "issue:45071", "right": "issue:46512", "accept": false, - "reason": "One is a type-checking regression in `PretrainedConfig`; the other is `from_config(dtype=...)` leaving `model.config.dtype` stale. Different APIs and symptoms." + "reason": "Type-checking regression in v5.4.0 is not the same as stale config.dtype after from_config(dtype=...)." }, { "left": "issue:42222", "right": "issue:46752", "accept": false, - "reason": "VitPose model breakage and empty `apply_chat_template` input crash are unrelated." + "reason": "Vitpose model breakage and empty chat-template handling are unrelated." }, { "left": "issue:44263", "right": "issue:46762", "accept": false, - "reason": "`torch.split()` return-value handling in an indexer is not the same bug as numerical drift in Minimax MSA attention." + "reason": "torch.split return semantics in an indexer is not the same as numerical drift in Minimax attention." }, { "left": "issue:44877", "right": "issue:46231", "accept": false, - "reason": "Strict config loading for `granite_speech` is unrelated to a documentation mismatch in `set_training()`." + "reason": "Strict config loading and a documentation parameter-name typo are not the same defect." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different tokenizers and different outputs (`offset_mapping` vs incorrect tokenization behavior)." + "reason": "Both are tokenizer-related, but they affect different models and different failure modes." }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "Flash-attention training degeneration is unrelated to a `save_pretrained` checkpoint regression for Qwen3.5." + "reason": "FlashAttention training degeneration and save_pretrained checkpoint corruption are separate issues." }, { "left": "issue:45632", "right": "issue:46981", "accept": false, - "reason": "A local-model cache-path collision and repeated 429 retries in `AutoProcessor` are distinct download/cache behaviors." + "reason": "Cache-path collision for local models is unrelated to repeated retries on HTTP 429 responses." }, { "left": "issue:45341", "right": "issue:45561", "accept": false, - "reason": "A small bug in `testing_utils.py` is not the same as a pytest-xdist race on `captured_info.txt`." + "reason": "A small testing_utils bug and an xdist race on captured_info.txt are not the same underlying issue." }, { "left": "issue:44410", "right": "issue:45310", "accept": false, - "reason": "Missing projections in qwen3next layer 0 and a Qwen3.5 MoE `from_pretrained` failure are different model-architecture/loading issues." + "reason": "Missing Qwen3Next projections and Qwen3.5 MoE from_pretrained failure are different model-loading problems." }, { "left": "issue:45071", "right": "issue:47059", "accept": false, - "reason": "Type checking for `PretrainedConfig` and a CLI completion error from `huggingface_hub` are unrelated." + "reason": "PretrainedConfig type checking and a CLI completion attribute error are unrelated." }, { - "left": "issue:45440", + "left": "issue:45341", "right": "issue:46328", "accept": false, - "reason": "Divergence from a reference implementation and a padding performance regression are different kinds of problems." + "reason": "Testing-utils bug versus a pad performance inefficiency; same file area, different bug." }, { "left": "issue:39401", "right": "issue:43421", "accept": false, - "reason": "Wrong tokenizer `offset_mapping` and runtime post-processor updates are separate tokenizer issues with different mechanics." + "reason": "Wrong offset_mapping and runtime post-processor updates are different tokenizer behaviors." }, { "left": "issue:44483", "right": "issue:45464", "accept": false, - "reason": "Both mention chat/completions, but one is an acceptance failure and the other is a streaming inference failure; not the same concrete bug." + "reason": "Both mention chat/completions, but one is request acceptance and the other is streaming inference failure on a specific model." }, { - "left": "issue:44704", - "right": "issue:47110", + "left": "issue:38617", + "right": "issue:41628", "accept": false, - "reason": "Passing kwargs to `cached_file` in `AutoProcessor.from_pretrained` is unrelated to `special_tokens` being a list instead of a dict." + "reason": "Different missing imports: layer_type_validation vs AutoImageProcessor." }, { - "left": "issue:43937", - "right": "issue:44496", + "left": "issue:38617", + "right": "issue:44220", "accept": false, - "reason": "Invalid `GenerationConfig` and unrecognized model/config loading are distinct loading/validation failures." + "reason": "ImportError for a config utility is unrelated to a feature-extraction function issue." }, { - "left": "issue:38617", - "right": "issue:41628", + "left": "issue:43937", + "right": "issue:44496", "accept": false, - "reason": "Both are import errors, but they concern different symbols and modules (`layer_type_validation` vs `AutoImageProcessor`)." + "reason": "Invalid GenerationConfig and missing model_type in config.json are distinct loading/configuration errors." }, { - "left": "issue:38617", - "right": "issue:44220", + "left": "issue:44704", + "right": "issue:47110", "accept": false, - "reason": "Importing a missing config symbol is unrelated to the `_torch_extract_fbank_features()` issue." + "reason": "Forwarding kwargs to cached_file and special_tokens list handling are not the same code-path problem." }, { "left": "issue:43122", "right": "issue:46298", "accept": false, - "reason": "Tokenization changes and a noisy logging message do not describe the same underlying bug." + "reason": "A tokenizer regression is not a noisy logging message bug." }, { "left": "issue:44964", "right": "issue:45278", "accept": false, - "reason": "A specific multimodal model load failure and a broad report of many import errors after upgrade overlap only at a high level; not the same concrete issue." + "reason": "A multimodal model load failure and a broad import-error report overlap only at the surface level." }, { "left": "issue:43097", "right": "issue:44493", "accept": false, - "reason": "Removed `tie_embeddings_and_encoder_decoder` support and unexpected position-id keys are unrelated regression reports." - }, - { - "left": "issue:42371", - "right": "issue:44261", - "accept": false, - "reason": "TF32 behavior guidance and a missing `rms_norm_eps` in MLA are unrelated." - }, - { - "left": "issue:42371", - "right": "issue:42831", - "accept": false, - "reason": "TF32 API guidance and FineGrainedFP8 accuracy problems are different topics." - }, - { - "left": "issue:44261", - "right": "issue:46762", - "accept": false, - "reason": "A layer-norm epsilon config bug is not the same as attention numerical discrepancy." - }, - { - "left": "issue:43208", - "right": "issue:46897", - "accept": false, - "reason": "xLSTM training blockers and Florence2 double-shift loss bug are unrelated model-specific training issues." - }, - { - "left": "issue:44484", - "right": "issue:45310", - "accept": false, - "reason": "`save_pretrained` shard-size behavior and Qwen3.5 MoE `from_pretrained` failure are different code paths." - }, - { - "left": "issue:44077", - "right": "issue:46981", - "accept": false, - "reason": "`patchtsmixer` post-init semantics and `AutoProcessor` HTTP 429 retry behavior are unrelated." - }, - { - "left": "issue:43931", - "right": "issue:44410", - "accept": false, - "reason": "Model weight-shape mismatch for Qwen3-VL and missing projections in qwen3next are not the same bug." + "reason": "Removed config field and unexpected position-id key errors are different regression types." } ] }, @@ -17943,9 +17831,10 @@ "cluster_id": "cluster-10000-540", "nodes": [ "issue:27957", - "issue:29159", "issue:29925", "issue:30947", + "issue:42371", + "issue:42831", "issue:43012", "issue:43066", "issue:43097", @@ -17955,17 +17844,15 @@ "issue:43441", "issue:43452", "issue:43531", - "issue:43550", "issue:43577", "issue:43582", "issue:43673", - "issue:43873", "issue:43874", "issue:43931", "issue:43957", - "issue:43986", "issue:44038", "issue:44075", + "issue:44077", "issue:44117", "issue:44230", "issue:44261", @@ -17978,20 +17865,28 @@ "issue:44829", "issue:44987", "issue:45071", + "issue:45310", "issue:45341", - "issue:45372", "issue:45478", "issue:45542", "issue:45588", "issue:45656", - "issue:45753", "issue:46050", "issue:46172", - "issue:46369", "issue:46752", + "issue:46762", + "issue:46897", + "issue:46981", "issue:47116" ], "soft_pairs": [ + "issue:42371|issue:44261", + "issue:42371|issue:42831", + "issue:44261|issue:46762", + "issue:43208|issue:46897", + "issue:44077|issue:46981", + "issue:43931|issue:44410", + "issue:44484|issue:45310", "issue:44075|issue:45656", "issue:43066|issue:43122", "issue:44410|issue:45478", @@ -18006,238 +17901,231 @@ "issue:43012|issue:43577", "issue:43957|issue:44351", "issue:43874|issue:44261", - "issue:46752|issue:47116", "issue:43531|issue:45341", + "issue:46752|issue:47116", + "issue:43097|issue:45588", "issue:27957|issue:43352", "issue:43582|issue:45341", - "issue:43097|issue:45588", "issue:44117|issue:44987", "issue:44263|issue:46172", "issue:44484|issue:45478", "issue:44360|issue:46172", - "issue:43097|issue:46050", - "issue:43986|issue:45372", - "issue:44261|issue:44485", - "issue:29159|issue:30947", - "issue:43097|issue:43873", - "issue:43208|issue:43550", - "issue:45753|issue:46369", - "issue:43986|issue:45341" + "issue:43097|issue:46050" ], - "prepared_review_unit_hash": "1e96d571350fdc3506e6b2e6cd371b612c25690e397961f1192daf840572a3cb", + "prepared_review_unit_hash": "cebcd8f15b839a35eadc69cbc36b9fb15f9e3ad25c988b0c8d9f412b6b04a57a", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36571, - "estimated_input_tokens": 9143, - "estimated_eval_tokens": 18542 + "serialized_chars": 36566, + "estimated_input_tokens": 9142, + "estimated_eval_tokens": 18540 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: mostly unrelated Transformers v5 regressions across tokenizers, model loading, FlashAttention, and model-specific code paths. I do not see any soft-edge pair that clearly reflects the same underlying bug or fix.", - "confidence": 0.9, - "canonical_issue_reason": "Issue 43122 is the most representative broad regression report here, since it describes cross-version tokenization behavior change in a general way rather than a narrowly model-specific failure.", + "summary": "No soft pair looks like a true duplicate; the candidates are mostly distinct regressions or model-specific bugs across unrelated code paths. The strongest standalone issue is the v5 API-removal report because it is concrete, well-scoped, and broadly actionable.", + "confidence": 0.67, + "canonical_issue_reason": "issue:43097 is the clearest, most actionable report: a specific v5 regression about an API being removed without notice, with direct user impact and a well-defined fix target.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43122 is the best global representative because it captures a user-visible version regression affecting tokenizer output, which is closer to the common theme than the many model-specific tickets.", + "best_issue_reason": "issue:43097 is the best single issue to anchor this set because it is concrete, broadly relevant to v5 migration, and easier to triage/fix than the more model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API-settings guidance is unrelated to the MLA q_a_layernorm rms_norm_eps precision bug." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "TF32 behavior control and FineGrainedFP8 accuracy are different numeric issues with different code paths." + }, + { + "left": "issue:44261", + "right": "issue:46762", + "accept": false, + "reason": "Both mention numerical differences, but one is a missing eps/config bug and the other is a Minimax attention accuracy mismatch." + }, + { + "left": "issue:43208", + "right": "issue:46897", + "accept": false, + "reason": "xLSTM training blockers and Florence2 double-shift loss are unrelated model-specific bugs." + }, + { + "left": "issue:44077", + "right": "issue:46981", + "accept": false, + "reason": "Optional post_init validation and HF_ENDPOINT retry behavior are different subsystems with no shared bug path." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "A Qwen3-VL load shape mismatch is not the same as missing attn_qkv/attn_gate projections in qwen3next." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "Default shard-size behavior is unrelated to the Qwen3.5 MoE from_pretrained failure." + }, { "left": "issue:44075", "right": "issue:45656", "accept": false, - "reason": "Both are optimizer-related, but one is SGD args being ignored while the other is a double-step issue with DeepSpeed; different code paths and failures." + "reason": "SGD argument handling and deepspeed causing double optimizer steps are different optimizer bugs." }, { "left": "issue:43066", "right": "issue:43122", "accept": false, - "reason": "Both mention tokenizer regressions in v5, but one is a wrong decoder type and the other is changed tokenization output; not the same bug." + "reason": "Wrong tokenizer decoder type and changed tokenization output are related only at a high level, not the same concrete bug." }, { "left": "issue:44410", "right": "issue:45478", "accept": false, - "reason": "Both are Qwen-family model loading issues, but they affect different model variants and different failure modes." + "reason": "Missing projections in qwen3next and Qwen3.5 MoE from_pretrained errors affect different model families and failure modes." }, { "left": "issue:44263", "right": "issue:44485", "accept": false, - "reason": "Both are GLM-related, but one is a torch.split indexing issue and the other is a RoPE implementation discussion; distinct problems." + "reason": "A torch.split return-value issue in the DSA indexer is not the same as the GLM-5 RoPE implementation bug." }, { "left": "issue:29925", "right": "issue:30947", "accept": false, - "reason": "Both are tokenizer bugs, but one is about enforcing use_fast and the other is BOS/EOS tokens not being applied; separate behaviors." + "reason": "SigLIP fast-tokenizer enforcement and LLaMA3 BOS/EOS insertion are separate tokenizer bugs." }, { "left": "issue:43673", "right": "issue:45341", "accept": false, - "reason": "A GenerationMixin cache regression and a testing_utils bug are unrelated." + "reason": "Chunked-prefill cache handling and a testing_utils bug have no shared code path." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "One is a removed config option in v5, the other is a gguf_file/from_pretrained loading failure; different underlying issues." + "reason": "An API removal warning and a gguf_file loading breakage are different from_pretrained issues, not one duplicate bug." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Both involve Qwen VL models, but the reported mismatches and loading failures are not the same code-path problem." + "reason": "Both are Qwen3-VL-related, but one is a weight-shape loading mismatch and the other is a broader v5/MoE bug report." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "PretrainedConfig type checking and missing TensorFlow backend handling are unrelated." + "reason": "PretrainedConfig type checking and the tensorboard-without-TensorFlow backend error are unrelated regressions." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "One is a bug in Qwen3-VL-Moe loading; the other is an fp8 support request. Not duplicates." + "reason": "A generic Qwen3-VL-Moe v5 bug and fp8 inference support are not the same fixable change." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "Both mention FlashAttention, but one is a model support/load failure and the other is degenerate training with flash_attention_3; different problems." + "reason": "FlashAttention model-load failure and flash_attention_3 degenerate training are distinct failures in different code paths." }, { "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "A compile-time precision warning and incorrect dtype propagation on Blip2 are not the same issue." + "reason": "The bfloat16 compile warning and BLIP2 dtype remaining float32 are different dtype-handling bugs." }, { "left": "issue:43957", "right": "issue:44351", "accept": false, - "reason": "Meta-device model loading breakage and a missing HybridCache import are unrelated." + "reason": "Meta-device model loading breakage and missing HybridCache import are unrelated v5 API/runtime issues." }, { "left": "issue:43874", "right": "issue:44261", "accept": false, - "reason": "Missing image processor method and missing rms_norm_eps in MLA q_a_layernorm are different bugs." + "reason": "Missing image-processor method and MLA q_a_layernorm precision error are different multimodal vs numeric bugs." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and a testing_utils bug are unrelated." }, { "left": "issue:46752", "right": "issue:47116", "accept": false, - "reason": "An empty-list crash in apply_chat_template and a streaming pipeline regression are not the same failure." + "reason": "Empty-list chat-template IndexError and streaming pipeline regression are separate failures." }, { - "left": "issue:43531", - "right": "issue:45341", + "left": "issue:43097", + "right": "issue:45588", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and a testing_utils bug are unrelated." + "reason": "API-removal visibility and flash_attention.py crashing on s_aux=None are different issues in different parts of the stack." }, { "left": "issue:27957", "right": "issue:43352", "accept": false, - "reason": "Both concern FlashAttention 2 support, but for different models and different support gaps." + "reason": "Flash Attention 2 support for XLMRoberta and NemotronH are model-specific support requests, not duplicates." }, { "left": "issue:43582", "right": "issue:45341", "accept": false, - "reason": "Apple Silicon caching allocator TypeError and a testing_utils bug are unrelated." - }, - { - "left": "issue:43097", - "right": "issue:45588", - "accept": false, - "reason": "A removed v5 config field and a flash_attention.py crash on s_aux=None are different code paths." + "reason": "AppleSilicon allocator warmup TypeError and testing_utils are unrelated." }, { "left": "issue:44117", "right": "issue:44987", "accept": false, - "reason": "Tokenizer mapping None handling and a specific model loading failure are not the same underlying bug." + "reason": "TOKENIZER_MAPPING_NAMES returning None and a model-loading failure in transformers>=5.1.0 are different from_pretrained problems." }, { "left": "issue:44263", "right": "issue:46172", "accept": false, - "reason": "Different tensor-shape/manipulation bugs in different model families; no shared concrete fix." + "reason": "GlmMoeDsaIndexer torch.split behavior and deepseek-v4 comb transpose logic are unrelated implementation bugs." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "A save_pretrained shard-size question and a Qwen3.5 MoE loading error are unrelated." + "reason": "max_shard_size default discussion and the Qwen3.5 MoE from_pretrained error are not the same issue." }, { "left": "issue:44360", "right": "issue:46172", "accept": false, - "reason": "A DSA indexer activation issue and a DeepSeek tensor operation issue are unrelated." + "reason": "DSA missing ReLU and deepseek-v4 transpose/dtype behavior are different code paths." }, { "left": "issue:43097", "right": "issue:46050", "accept": false, - "reason": "Removed encoder-decoder tying and quantization save/load breakage are different regressions." - }, - { - "left": "issue:43986", - "right": "issue:45372", - "accept": false, - "reason": "Missing torchvision and missing ReasoningEffort imports are separate dependency/import problems." - }, - { - "left": "issue:44261", - "right": "issue:44485", - "accept": false, - "reason": "Both are GLM-related, but one is a missing config epsilon issue and the other is a RoPE implementation discussion." - }, - { - "left": "issue:29159", - "right": "issue:30947", - "accept": false, - "reason": "General slow-vs-fast tokenizer inconsistency and LLaMA3 BOS/EOS token application are different tokenizer bugs." - }, - { - "left": "issue:43097", - "right": "issue:43873", - "accept": false, - "reason": "A removed v5 field and offloading/quantization behavior are not the same bug." - }, - { - "left": "issue:43208", - "right": "issue:43550", - "accept": false, - "reason": "An xLSTM training triad of bugs and a Bamba SDPA torch.compile failure are separate model-specific issues." - }, - { - "left": "issue:45753", - "right": "issue:46369", - "accept": false, - "reason": "An infinite loop on a specific image and a memory overflow on sam3-litetext are different multimodal failures." - }, - { - "left": "issue:43986", - "right": "issue:45341", - "accept": false, - "reason": "A processor import crash without torchvision and a testing_utils bug are unrelated." + "reason": "An API removal warning and a sinq save/load quantization breakage are unrelated." } ] }, @@ -18261,6 +18149,7 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ + "issue:29159", "issue:30947", "issue:33453", "issue:34567", @@ -18270,53 +18159,58 @@ "issue:42890", "issue:43012", "issue:43064", + "issue:43097", "issue:43208", - "issue:43502", - "issue:43526", "issue:43550", "issue:43673", + "issue:43873", "issue:43874", "issue:43986", - "issue:43992", "issue:44038", "issue:44062", "issue:44117", "issue:44246", + "issue:44261", "issue:44263", "issue:44315", - "issue:44380", "issue:44410", - "issue:44451", "issue:44484", + "issue:44485", "issue:44704", "issue:44936", - "issue:44938", "issue:45103", + "issue:45341", + "issue:45372", "issue:45375", "issue:45440", - "issue:45712", + "issue:45753", "issue:46123", "issue:46172", - "issue:46178", "issue:46298", "issue:46365", - "issue:46421", + "issue:46369", "issue:46655", "issue:46664", "issue:46701", "issue:46710", "issue:46752", "issue:46821", - "issue:46981", - "issue:47116" + "issue:46981" ], "soft_pairs": [ + "issue:43986|issue:45372", + "issue:44261|issue:44485", + "issue:29159|issue:30947", + "issue:43097|issue:43873", + "issue:43208|issue:43550", + "issue:43986|issue:45341", + "issue:45753|issue:46369", "issue:45440|issue:46172", "issue:44246|issue:46981", "issue:44117|issue:46701", - "issue:44062|issue:44117", "issue:43012|issue:46821", "issue:44410|issue:45375", + "issue:44062|issue:44117", "issue:30947|issue:33453", "issue:43673|issue:46123", "issue:30947|issue:34567", @@ -18325,224 +18219,210 @@ "issue:46655|issue:46664", "issue:46172|issue:46710", "issue:46298|issue:46752", - "issue:44936|issue:46365", - "issue:43986|issue:44484", "issue:38617|issue:42757", "issue:44704|issue:45103", + "issue:44936|issue:46365", "issue:43064|issue:44315", "issue:38617|issue:41950", - "issue:42890|issue:46298", - "issue:43502|issue:43992", - "issue:44380|issue:46421", - "issue:44451|issue:47116", - "issue:44938|issue:46178", - "issue:43208|issue:43526", - "issue:43064|issue:47116", - "issue:43012|issue:43550", - "issue:45712|issue:46298" + "issue:43986|issue:44484", + "issue:42890|issue:46298" ], - "prepared_review_unit_hash": "748c3661d87195c4c07dd4528bfb0c71a0d047a7d7d046b6e0ff41fccad35e0d", + "prepared_review_unit_hash": "afe1983dae7fc6d4d969f7af58e7d9b2e92b0cc05e73f71f58ac3506e39d9b95", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36209, - "estimated_input_tokens": 9053, - "estimated_eval_tokens": 18362 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35286, + "estimated_input_tokens": 8822, + "estimated_eval_tokens": 17900 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is heterogeneous: the soft pairs mostly share only a broad subsystem name or generic regression wording, not the same concrete bug. I would reject all soft-edge duplicate candidates and avoid selecting a single canonical issue/PR for the whole set.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "This is a loose semantic cluster of mostly unrelated Transformers issues. The soft edges mostly connect items that share a subsystem name (tokenizers, processors, DeepSeek/Qwen variants, trainer, etc.) but not the same concrete bug or fix path, so none should be merged as duplicates.", + "confidence": 0.62, + "canonical_issue_reason": "Issue 33453 is the broadest representative of the cluster: a tokenizer-loading regression with wording that sits nearest to the other tokenizer-related reports. It is still not a duplicate of most others, but it is the best single anchor issue for this mixed cluster.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "33453 is the best global issue candidate because it is generic, central to one of the recurring themes in the cluster, and broad enough to serve as the representative issue even though the cluster is not actually a tight duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Both are processor-loading import crashes, but they come from different missing dependencies and different code paths (video model/torchvision vs Gemma 4/mistral_common)." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "These describe different bugs: a q_a_layernorm precision/config issue versus a GLM-5 RoPE implementation question." + }, + { + "left": "issue:29159", + "right": "issue:30947", + "accept": false, + "reason": "Both are tokenizer-related, but one is slow/fast tokenizer inconsistency and the other is LLaMA3 special-token flags not taking effect; not the same concrete bug." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "One is about a removed model API flag, the other about offloading with quantization; different features and failure modes." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs and a Bamba torch.compile/SDPA failure are unrelated model-specific problems." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "A missing torchvision import during processor loading is unrelated to a small testing_utils bug." + }, + { + "left": "issue:45753", + "right": "issue:46369", + "accept": false, + "reason": "These affect different models and symptoms: an infinite loop on one image versus a memory overflow in sam3-litetext." + }, { "left": "issue:45440", "right": "issue:46172", "accept": false, - "reason": "Different DeepSeek-related failures: model divergence vs a dtype/transpose bug in a separate MoE path." + "reason": "Both mention DeepSeek, but one is a native-vs-remote implementation divergence and the other is a dtype/transpose bug in a different model path." }, { "left": "issue:44246", "right": "issue:46981", "accept": false, - "reason": "Both involve loading/network behavior, but one is slow imports while the other is repeated 429 retries in AutoProcessor." + "reason": "Import slowness is not the same as repeated 429 retry behavior in AutoProcessor with HF_ENDPOINT." }, { "left": "issue:44117", "right": "issue:46701", "accept": false, - "reason": "Unrelated failures: tokenizer mapping fallback vs an image-text-to-text pipeline crash." - }, - { - "left": "issue:44062", - "right": "issue:44117", - "accept": false, - "reason": "One is an AddedToken constructor argument conflict; the other is tokenizer mapping returning None." + "reason": "A TOKENIZER_MAPPING_NAMES None-handling bug is unrelated to the DiffusionGemma image-text pipeline failure." }, { "left": "issue:43012", "right": "issue:46821", "accept": false, - "reason": "Different bugs and code paths: PyTorch bfloat16 warning vs a TimesFM window_size AttributeError." + "reason": "These are different issues: a PyTorch precision warning during compilation versus an AttributeError on TimesFM window_size." }, { "left": "issue:44410", "right": "issue:45375", "accept": false, - "reason": "Both are Qwen3-related, but one is missing projections in layer 0 and the other is a config field being dropped." + "reason": "Missing projections in Qwen3Next and a silently dropped config field in Qwen3_5MoeVisionConfig are different model/config bugs." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "AddedToken keyword duplication and TOKENIZER_MAPPING_NAMES returning None are different tokenizer-layer failures." }, { "left": "issue:30947", "right": "issue:33453", "accept": false, - "reason": "Tokenizer-related, but one is about BOS/EOS flags not being applied and the other is a broader tokenizer loading regression." + "reason": "The first is a LLaMA3 add_bos/add_eos behavior issue; the second is a broader tokenizer-loading regression. Related area, but not the same bug." }, { "left": "issue:43673", "right": "issue:46123", "accept": false, - "reason": "Both touch generation/prefill behavior, but cache initialization and final-batch masking are different bugs." + "reason": "Generation cache missing in chunked_prefill and mask pipeline batch-finalization are different pipeline logic bugs." }, { "left": "issue:30947", "right": "issue:34567", "accept": false, - "reason": "Completely different areas: tokenizer special-token handling vs TrainerState token counting." + "reason": "Tokenizer special-token behavior and TrainerState token counting are unrelated." }, { "left": "issue:43874", "right": "issue:44263", "accept": false, - "reason": "Different components and symptoms: image processor method missing vs incorrect torch.split return handling." + "reason": "Missing get_number_of_image_patches and torch.split return values are different GLM-related issues with different failing code paths." }, { "left": "issue:44038", "right": "issue:44410", "accept": false, - "reason": "Both mention Qwen3-VL/Qwen3 family, but one is a Transformers v5 compatibility issue and the other is a missing projection bug." + "reason": "Qwen3-VL-Moe and Qwen3Next are different model families with different bugs." }, { "left": "issue:46655", "right": "issue:46664", "accept": false, - "reason": "Same model family, but padding calculation and num_frames semantics are distinct video-processor bugs." + "reason": "Both are SmolVLM video-processor issues, but one is padding calculation and the other is num_frames semantics; they are distinct fixes." }, { "left": "issue:46172", "right": "issue:46710", "accept": false, - "reason": "Different DeepSeek issues: a low-level tensor operation bug vs an accuracy regression after upgrade." + "reason": "A tensor-operation bug in deepseek-v4 is not the same as an accuracy regression after upgrading Transformers for a different model." }, { "left": "issue:46298", "right": "issue:46752", "accept": false, - "reason": "No overlap beyond general API area; noisy logging and empty-list chat templating are unrelated." - }, - { - "left": "issue:44936", - "right": "issue:46365", - "accept": false, - "reason": "Trainer evaluation failure after train is unrelated to an import conflict with datasets." - }, - { - "left": "issue:43986", - "right": "issue:44484", - "accept": false, - "reason": "Video-model AutoProcessor crash without torchvision is unrelated to save_pretrained shard size defaults." + "reason": "Noisy logging and an empty-list apply_chat_template crash are unrelated." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are import errors from dependency churn, but they concern different symbols and different upstream packages." + "reason": "These are different import errors from different external packages and symbols; not the same failure." }, { "left": "issue:44704", "right": "issue:45103", "accept": false, - "reason": "AutoProcessor kwarg forwarding and auto_docstring annotation handling are separate code paths." + "reason": "AutoProcessor kwargs forwarding and an auto_docstring annotations crash are unrelated code paths." + }, + { + "left": "issue:44936", + "right": "issue:46365", + "accept": false, + "reason": "trainer.evaluate() after train() and Trainer import blocked by datasets are different trainer-related problems." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "Trainer/FSDP/PEFT optimizer-state corruption is unrelated to Liger Kernel not being applied with model_init." + "reason": "A distributed optimizer-state bug and Liger Kernel not being applied with model_init are different training/integration issues." }, { "left": "issue:38617", "right": "issue:41950", "accept": false, - "reason": "An import error for a removed utility is not the same as video-classification pipeline processor selection." - }, - { - "left": "issue:42890", - "right": "issue:46298", - "accept": false, - "reason": "Test nondeterminism from missing set_seed is unrelated to a logging-message issue." - }, - { - "left": "issue:43502", - "right": "issue:43992", - "accept": false, - "reason": "local_files_only network leakage and a missing embed_tokens weight in UMT5Encoder are different loading bugs." - }, - { - "left": "issue:44380", - "right": "issue:46421", - "accept": false, - "reason": "Attention scaling being ignored and cache lazy-init recompilation are separate generation/runtime problems." - }, - { - "left": "issue:44451", - "right": "issue:47116", - "accept": false, - "reason": "Model loading failure for ScandiBERT is unrelated to a streaming pipeline regression." - }, - { - "left": "issue:44938", - "right": "issue:46178", - "accept": false, - "reason": "Python 3.14 import failure and an invalid backend name in BACKENDS_MAPPING are different import/config issues." - }, - { - "left": "issue:43208", - "right": "issue:43526", - "accept": false, - "reason": "xLSTM training blockers and BeitImageProcessorFast label reduction are unrelated." - }, - { - "left": "issue:43064", - "right": "issue:47116", - "accept": false, - "reason": "Optimizer-state corruption in training is unrelated to the streaming pipeline regression." + "reason": "An import error for layer_type_validation and a video-classification pipeline looking for image processors are unrelated." }, { - "left": "issue:43012", - "right": "issue:43550", + "left": "issue:43986", + "right": "issue:44484", "accept": false, - "reason": "A bfloat16 compile warning and a Bamba SDPA compile failure are not the same bug." + "reason": "Processor-loading failures and a save_pretrained shard-size question are not the same issue." }, { - "left": "issue:45712", + "left": "issue:42890", "right": "issue:46298", "accept": false, - "reason": "Leftover dummy classes and noisy logging do not describe the same underlying change or failure." + "reason": "A flaky SAM HQ test due to missing set_seed and a noisy logging complaint are unrelated." } ] }, @@ -18573,17 +18453,18 @@ "issue:42175", "issue:42757", "issue:42831", - "issue:42890", "issue:42907", "issue:43012", "issue:43064", "issue:43066", "issue:43097", + "issue:43208", "issue:43352", + "issue:43502", + "issue:43526", + "issue:43550", "issue:43582", - "issue:43618", "issue:43756", - "issue:43761", "issue:43824", "issue:43825", "issue:43827", @@ -18592,45 +18473,53 @@ "issue:43992", "issue:44077", "issue:44162", - "issue:44222", "issue:44261", "issue:44368", "issue:44380", - "issue:44483", + "issue:44451", "issue:44484", "issue:44485", - "issue:44514", "issue:44655", "issue:44734", - "issue:45478", + "issue:44938", "issue:45479", "issue:45584", + "issue:45712", "issue:45753", "issue:46132", "issue:46172", + "issue:46178", "issue:46298", "issue:46365", - "issue:47059", + "issue:46421", "issue:47116", "issue:47276" ], "soft_pairs": [ - "issue:43066|issue:43906", + "issue:43502|issue:43992", + "issue:44380|issue:46421", + "issue:44451|issue:47116", + "issue:44938|issue:46178", + "issue:43208|issue:43526", + "issue:43064|issue:47116", + "issue:43012|issue:43550", + "issue:45712|issue:46298", "issue:42907|issue:45753", + "issue:43066|issue:43906", "issue:42907|issue:44655", "issue:45584|issue:46365", "issue:43012|issue:44368", "issue:43825|issue:44655", "issue:44734|issue:47116", "issue:43827|issue:47116", - "issue:43992|issue:44484", "issue:43352|issue:43976", + "issue:43992|issue:44484", "issue:44077|issue:46365", "issue:30947|issue:33357", - "issue:44162|issue:46172", + "issue:44380|issue:46132", "issue:36683|issue:43582", "issue:44380|issue:45479", - "issue:44380|issue:46132", + "issue:44162|issue:46172", "issue:42757|issue:43824", "issue:42175|issue:46365", "issue:43012|issue:47276", @@ -18638,215 +18527,210 @@ "issue:43352|issue:44380", "issue:43097|issue:46298", "issue:36010|issue:42757", - "issue:42831|issue:44261", - "issue:44222|issue:45478", - "issue:44483|issue:44514", - "issue:42890|issue:47059", - "issue:43064|issue:44380", - "issue:43618|issue:43761" + "issue:42831|issue:44261" ], - "prepared_review_unit_hash": "42442e60f731a99d4cd7c94b8fb07933e148038c2b0f699b56cf7203e09f586d", + "prepared_review_unit_hash": "34b1ecfd656c4f4da239869e9bedc18e670e348ab8a3f5040fc62ce174f2c42d", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35659, - "estimated_input_tokens": 8915, - "estimated_eval_tokens": 18086 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37101, + "estimated_input_tokens": 9276, + "estimated_eval_tokens": 18808 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is a set of mostly unrelated false-positive similarities: import errors, model-specific regressions, save/load issues, warnings, docs, tests, and runtime bugs. I found no pair that clearly describes the same underlying change or bug.", - "confidence": 0.97, - "canonical_issue_reason": "No single issue cleanly represents a duplicate group here; the soft pairs span different subsystems and different failure modes.", + "summary": "These are mostly unrelated issues with only superficial topical overlap; none of the soft pairs look like true duplicates or the same change. If forced to pick a representative issue, issue 43097 is the clearest standalone regression report and has inbound references, but this set is too diverse to treat as one cluster.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 43097 is the most central, well-scoped regression report in this set and has inbound references; it is the best single representative only because the rest are clearly separate topics.", "canonical_pr_reason": null, - "best_issue_reason": "No strong global best issue: there is no real duplicate cluster to anchor, so any issue would be an arbitrary representative.", + "best_issue_reason": "Issue 43097 has the strongest triage value as a concrete, user-visible v5 regression with broader discussion/reference signals than the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43066", - "right": "issue:43906", + "left": "issue:43502", + "right": "issue:43992", "accept": false, - "reason": "Both mention transformer internals, but one is a tokenizer/type regression and the other is an isolated reproduction of a different issue." + "reason": "Different failures: network/API access despite local_files_only vs missing embed_tokens.weight in from_pretrained." }, { - "left": "issue:42907", - "right": "issue:45753", + "left": "issue:44380", + "right": "issue:46421", "accept": false, - "reason": "Save/dequantization failure and image-driven infinite loop are different bugs in unrelated code paths." + "reason": "GPT2 attention scaling being ignored is unrelated to cache lazy initialization causing recompilation." }, { - "left": "issue:42907", - "right": "issue:44655", + "left": "issue:44451", + "right": "issue:47116", "accept": false, - "reason": "Both involve saving, but one is about dequantized model save failure and the other about saving Pipeline objects; not the same bug." + "reason": "Model loading failure for a specific checkpoint is not the same bug as streaming pipeline regression in v5." }, { - "left": "issue:45584", - "right": "issue:46365", + "left": "issue:44938", + "right": "issue:46178", "accept": false, - "reason": "Whisper generation failure and Trainer import conflict are unrelated." + "reason": "Python 3.14 import/load failure and tensorflow_text backend mapping error are different import-path problems." }, { - "left": "issue:43012", - "right": "issue:44368", + "left": "issue:43208", + "right": "issue:43526", "accept": false, - "reason": "One is a PyTorch compile warning for bfloat16; the other is a tie_word_embeddings warning in a different model workflow." + "reason": "xLSTM training bugs and BeitImageProcessorFast reduce_labels returning one label are unrelated." }, { - "left": "issue:43825", - "right": "issue:44655", + "left": "issue:43064", + "right": "issue:47116", "accept": false, - "reason": "Both are pipeline-related, but one is an error-message/docs issue and the other is a save_pretrained failure." + "reason": "FSDP2/PEFT optimizer state corruption is a different code path from streaming pipeline behavior." }, { - "left": "issue:44734", - "right": "issue:47116", + "left": "issue:43012", + "right": "issue:43550", "accept": false, - "reason": "Different server/runtime regressions: KV-cache continuation indexing vs streaming pipeline behavior." + "reason": "A bfloat16 precision warning is not the same as Bamba SDPA torch.compile failure." }, { - "left": "issue:43827", - "right": "issue:47116", + "left": "issue:45712", + "right": "issue:46298", "accept": false, - "reason": "Docs still referencing pipeline() is not the same as a streaming pipeline runtime regression." + "reason": "Dummy-class cleanup/check_repo leakage is unrelated to a noisy logging message." }, { - "left": "issue:43992", - "right": "issue:44484", + "left": "issue:42907", + "right": "issue:45753", "accept": false, - "reason": "Missing checkpoint weight on load and a max_shard_size question on save are different save/load concerns." + "reason": "Saving dequantized models and an infinite-loop multimodal generation bug are different problems." }, { - "left": "issue:43352", - "right": "issue:43976", + "left": "issue:43066", + "right": "issue:43906", "accept": false, - "reason": "Flash Attention support error and Python version compatibility are unrelated." + "reason": "Wrong tokenizer decoder type is not the same as an isolated reproduction of another issue." }, { - "left": "issue:44077", - "right": "issue:46365", + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline save_pretrained failure is a different saving path than dequantized Ministrals/Devstrals save issues." + }, + { + "left": "issue:45584", + "right": "issue:46365", + "accept": false, + "reason": "Whisper empty-transcription generation bug is unrelated to Trainer import failure caused by datasets." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "A compile-time float32 warning is not the same as a tie_word_embeddings warning." + }, + { + "left": "issue:43825", + "right": "issue:47116", + "accept": false, + "reason": "Incorrect translation-task error text is separate from the streaming pipeline regression." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Unsupported Flash Attention 2 for NemotronH is unrelated to Python 3.9/3.10 compatibility." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing embed_tokens.weight on load is not the same as asking why max_shard_size defaults to 50GB." + }, + { + "left": "issue:44077", + "right": "issue:46365", "accept": false, - "reason": "PatchTSMixer post_init validation and importing Trainer with datasets installed are unrelated." + "reason": "Optional post_init handling in patchtsmixer is unrelated to the Trainer import issue." }, { "left": "issue:30947", "right": "issue:33357", "accept": false, - "reason": "Tokenizer BOS/EOS handling and a CLIP Mac bus error are unrelated." + "reason": "Tokenizer BOS/EOS handling does not match a CLIP bus error on macOS." }, { - "left": "issue:44162", - "right": "issue:46172", + "left": "issue:44380", + "right": "issue:46132", "accept": false, - "reason": "ESM2 breakage and a DeepSeek tensor operation bug are different model-specific issues." + "reason": "Ignored GPT2 attention scaling and altered registered-function behavior are different bugs." }, { "left": "issue:36683", "right": "issue:43582", "accept": false, - "reason": "Gemma3 config missing vocab_size and an Apple Silicon TypeError in caching warmup are unrelated." + "reason": "Missing vocab_size on Gemma3Config is unrelated to an Apple Silicon TypeError in caching_allocator_warmup." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "Attention scaling config ignored and zero-loss classification behavior are different bugs." + "reason": "Attention scaling being ignored is not the same as degenerate zero loss in sequence classification." }, { - "left": "issue:44380", - "right": "issue:46132", + "left": "issue:44162", + "right": "issue:46172", "accept": false, - "reason": "Both touch attention code, but one is a GPT2 scaling regression and the other changes AttentionInterface.register behavior." + "reason": "ESM2 breakage and deepseek-v4 tensor transpose behavior are separate model-specific issues." }, { "left": "issue:42757", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols from different packages." + "reason": "Two different import errors for different symbols and packages; not the same underlying bug." }, { "left": "issue:42175", "right": "issue:46365", "accept": false, - "reason": "Backend selection for pip extras and a Trainer import issue are unrelated dependency problems." + "reason": "Missing TensorFlow backend after pip install .[torch] is unrelated to Trainer import failure with datasets installed." }, { "left": "issue:43012", "right": "issue:47276", "accept": false, - "reason": "A compile-time precision warning and a synced_gpus warning message are not the same bug." + "reason": "A precision warning during compile and a misleading synced_gpus warning are different messages/issues." }, { "left": "issue:43756", "right": "issue:44485", "accept": false, - "reason": "Both mention RoPE, but they are model-specific implementation discrepancies, not the same underlying issue." - }, - { - "left": "issue:43352", - "right": "issue:44380", - "accept": false, - "reason": "Different features and failures: Flash Attention support gating vs attention scaling behavior." + "reason": "Smollm3 RoPE layer dropping and GLM-5 RoPE implementation discussion are not the same concrete bug." }, { "left": "issue:43097", "right": "issue:46298", "accept": false, - "reason": "Removed API without indication and a noisy logging message are unrelated." + "reason": "An API removal regression is unrelated to a noisy logging message." }, { "left": "issue:36010", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they are for different symbols and different dependency breakages." + "reason": "Importing GenerationMixin and importing is_offline_mode are different missing-symbol regressions." }, { "left": "issue:42831", "right": "issue:44261", "accept": false, - "reason": "Both concern precision/accuracy, but they describe different models and different numerical issues." - }, - { - "left": "issue:44222", - "right": "issue:45478", - "accept": false, - "reason": "MoE save_pretrained failure and MoE from_pretrained error are related thematically but not the same concrete bug." - }, - { - "left": "issue:44483", - "right": "issue:44514", - "accept": false, - "reason": "Chat completions endpoint behavior and Qwen2.5-VL chat template batching crash are unrelated." - }, - { - "left": "issue:42890", - "right": "issue:47059", - "accept": false, - "reason": "A flaky model integration test and a CLI compatibility break with huggingface_hub are unrelated test failures." - }, - { - "left": "issue:43064", - "right": "issue:44380", - "accept": false, - "reason": "FSDP2/PEFT optimizer state corruption and GPT2 attention scaling are entirely different code paths." - }, - { - "left": "issue:43618", - "right": "issue:43761", - "accept": false, - "reason": "Both are CLIP regressions, but attentions being unset and hidden_states missing are different outputs and likely different fixes." + "reason": "FineGrainedFP8 accuracy issues and MLA q_a_layernorm rms_norm_eps precision errors are distinct." } ] }, @@ -18876,24 +18760,26 @@ "issue:36296", "issue:36683", "issue:38617", - "issue:42548", - "issue:43010", + "issue:42890", + "issue:43064", "issue:43299", "issue:43352", "issue:43475", "issue:43502", "issue:43611", "issue:43618", + "issue:43761", "issue:43867", "issue:43976", "issue:43992", - "issue:44075", "issue:44077", + "issue:44222", "issue:44263", - "issue:44351", "issue:44355", "issue:44380", "issue:44393", + "issue:44483", + "issue:44514", "issue:44625", "issue:44655", "issue:44855", @@ -18901,8 +18787,7 @@ "issue:44938", "issue:45216", "issue:45276", - "issue:45278", - "issue:45310", + "issue:45478", "issue:45593", "issue:45759", "issue:46077", @@ -18911,7 +18796,6 @@ "issue:46512", "issue:46851", "issue:46877", - "issue:47000", "issue:47059", "issue:47110", "issue:47317", @@ -18920,6 +18804,11 @@ "issue:47337" ], "soft_pairs": [ + "issue:44222|issue:45478", + "issue:44483|issue:44514", + "issue:43064|issue:44380", + "issue:42890|issue:47059", + "issue:43618|issue:43761", "issue:45276|issue:46077", "issue:44625|issue:46365", "issue:38617|issue:43502", @@ -18942,251 +18831,239 @@ "issue:36683|issue:44625", "issue:43618|issue:45216", "issue:36683|issue:47110", - "issue:44355|issue:44855", "issue:29870|issue:36296", + "issue:44355|issue:44855", "issue:43867|issue:46202", "issue:44908|issue:46851", "issue:43992|issue:47317", - "issue:44393|issue:47337", - "issue:44075|issue:46365", - "issue:43010|issue:45278", - "issue:44393|issue:47000", - "issue:43992|issue:45310", - "issue:42548|issue:44351", - "issue:29870|issue:33453" + "issue:44393|issue:47337" ], - "prepared_review_unit_hash": "e1c608fcdd9e02c05d0d82f42f78bc93dd190d186468a0a49e2698173e6b8c3a", + "prepared_review_unit_hash": "1a6c064ff108596663784c1b900da04f6ce0c897d4301dc5dfcb4ea7ebc86b4a", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37426, - "estimated_input_tokens": 9357, - "estimated_eval_tokens": 18970 + "soft_pair_count": 33, + "serialized_chars": 37332, + "estimated_input_tokens": 9333, + "estimated_eval_tokens": 18922 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated Transformers issues; the textual similarity is driven by broad themes like import/load regressions, config propagation, or model-specific attribute bugs, but none of the soft pairs look like true duplicates of the same underlying fix.", - "confidence": 0.29, - "canonical_issue_reason": "issue:45278 is the broadest umbrella here ('Many import errors after update from 4.57.0 to 5.5.0') and best matches the cluster\u2019s recurring import/load-regression motif, even though the other issues are mostly distinct.", + "summary": "This cluster is a loose mix of mostly unrelated Transformers issues. The soft pairs share themes like loading, saving, CLIP, Gemma, or Python-version regressions, but they do not look like the same concrete bug/change.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43611 is the broadest representative here: a Transformers 5.0 loading regression affecting models with `base_model_prefix`, which is closer to the cluster\u2019s recurring load/save/config breakage theme than the more narrowly scoped reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best representative because it is the most general, most cluster-like problem statement among otherwise narrow, model-specific bugs.", + "best_issue_reason": "Same reason: it is the most general, central loading-regression issue among an otherwise heterogeneous set, so it is the best single issue to anchor the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5/MoE and FP8/save-load behavior, but one is about saving pretrained Moe weights and the other about `from_pretrained` loading failure; too different to merge." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Different layers and failures: chat-completions request validation vs batched processor chat template padding behavior." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption is unrelated to GPT2 attention scaling under SDPA/FlashAttention." + }, + { + "left": "issue:42890", + "right": "issue:47059", + "accept": false, + "reason": "A flaky test seeding issue is not the same as a CLI completion compatibility break." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both are CLIP v5 regressions, but one is missing attentions and the other is missing hidden_states; related area, not clearly the same bug." + }, { "left": "issue:45276", "right": "issue:46077", "accept": false, - "reason": "Both are Gemma4-related, but one is about resize_token_embeddings not updating outputs while the other is about bidirectional attention still building causal masks; different code paths." + "reason": "Gemma4 embedding resize behavior and Gemma4 bidirectional-attention masking are separate code paths." }, { "left": "issue:44625", "right": "issue:46365", "accept": false, - "reason": "Qwen3.5 num_labels propagation is unrelated to an import failure caused by datasets; no shared underlying bug." + "reason": "Qwen3.5 `num_labels` propagation and an import conflict with `datasets` are unrelated." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "Missing import of layer_type_validation and ignoring local_files_only are separate failures in different areas." + "reason": "An import error for `layer_type_validation` is unrelated to `local_files_only=True` still triggering network access." }, { "left": "issue:44625", "right": "issue:46077", "accept": false, - "reason": "Different models and different bugs: Qwen3.5 num_labels propagation vs Gemma4 attention-mask construction." + "reason": "Same broad Gemma/Qwen family style, but the bugs are distinct config-vs-attention issues." }, { "left": "issue:44655", "right": "issue:47321", "accept": false, - "reason": "Saving Pipeline objects and an AMD quark class not being updated are unrelated issues." + "reason": "Saving `Pipeline` objects and an AMD quark class update problem are different." }, { "left": "issue:44655", "right": "issue:47332", "accept": false, - "reason": "Pipeline save_pretrained failure and disk-offloaded Qwen3-VL save failure are distinct save/offload paths." + "reason": "Both mention saving, but one is pipeline serialization and the other is disk-offloaded Qwen3-VL saving." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Qwen3VL MoE loading regression and sorted state_dict load error are different load-time problems with different triggers." + "reason": "Qwen3VL MoE loading regression and a sorted-state-dict load failure are not the same concrete problem." }, { "left": "issue:43611", "right": "issue:44655", "accept": false, - "reason": "base_model_prefix loading breakage is not the same bug as Pipeline save_pretrained serialization." + "reason": "Model loading with `base_model_prefix` and pipeline save_pretrained are distinct issues." }, { "left": "issue:43618", "right": "issue:44380", "accept": false, - "reason": "CLIPOutput attentions assignment and GPT2 attention scaling under SDPA/FlashAttention are different attention-related regressions." + "reason": "CLIP output-field regression and GPT2 attention scaling under backend selection are unrelated bugs." }, { "left": "issue:36683", "right": "issue:43475", "accept": false, - "reason": "Both mention missing attributes, but Gemma3Config vocab_size and SAM3VisionEncoderOutput fpn_position_embeddings are unrelated objects and code paths." + "reason": "Missing `vocab_size` in Gemma3Config is unrelated to missing `fpn_position_embeddings` in SAM 3 Video." }, { "left": "issue:43352", "right": "issue:45593", "accept": false, - "reason": "FlashAttention 2 support in NemotronHForCausalLM and D-FINE auxiliary-loss behavior are unrelated model features." + "reason": "Flash Attention 2.0 support gating is not the same bug as D-FINE auxiliary-loss handling." }, { "left": "issue:36683", "right": "issue:44077", "accept": false, - "reason": "Gemma3Config missing vocab_size is unrelated to patchtsmixer post_init validation." + "reason": "A missing config attribute and the `patchtsmixer` optional `post_init` restriction are unrelated." }, { "left": "issue:46077", "right": "issue:46512", "accept": false, - "reason": "Bidirectional attention mask construction and stale config.dtype after from_config are different bugs." + "reason": "Bidirectional attention mask construction and stale `config.dtype` after `from_config(dtype=...)` are different issues." }, { "left": "issue:43976", "right": "issue:44938", "accept": false, - "reason": "Both are Python-version compatibility complaints, but they refer to different releases and failure modes." + "reason": "Both are Python-version compatibility reports, but they target different versions and different failures." }, { "left": "issue:44263", "right": "issue:45759", "accept": false, - "reason": "A torch.split return-value issue in GlmMoeDsaIndexer is unrelated to unwrapping text_config in composite Qwen configs." + "reason": "A `torch.split()` return-value issue in GlmMoeDsaIndexer is unrelated to Qwen composite-config unwrapping in `from_config`." }, { "left": "issue:44355", "right": "issue:46877", "accept": false, - "reason": "Compiled Python import errors and auto_docstring stdout diagnostics are different import-time problems." + "reason": "Compiled-Python-file errors and noisy auto-docstring lint output are unrelated." }, { "left": "issue:29925", "right": "issue:33453", "accept": false, - "reason": "SigLIP use_fast enforcement is too specific to be the same as the broader tokenizer-loading regression." + "reason": "Tokenizer fast-mode enforcement and a general tokenizer-loading regression are too broad to treat as the same bug." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "A load error when state_dict is sorted is not the same as a Qwen3.5 save_pretrained regression." + "reason": "Sorted state-dict loading errors and Qwen3.5 checkpoint saving corruption are different directions of failure." }, { "left": "issue:44938", "right": "issue:47059", "accept": false, - "reason": "Python 3.14 import/load failure and a huggingface_hub CLI compatibility issue are different environment regressions." + "reason": "Python 3.14 import/load failure and the huggingface_hub CLI completion regression are unrelated." }, { "left": "issue:36683", "right": "issue:44625", "accept": false, - "reason": "Gemma3Config vocab_size and Qwen3.5 num_labels propagation are separate configuration bugs." + "reason": "Gemma3 config `vocab_size` and Qwen3.5 `num_labels` propagation are separate config bugs." }, { "left": "issue:43618", "right": "issue:45216", "accept": false, - "reason": "CLIPOutput attentions missing and Qwen3.5 checkpoint saving corruption are unrelated regressions." + "reason": "CLIP attentions regression and Qwen3.5 checkpoint save corruption do not share the same code path." }, { "left": "issue:36683", "right": "issue:47110", "accept": false, - "reason": "Missing Gemma3Config vocab_size and special_tokens being a list instead of a dict are unrelated attribute/type errors." + "reason": "Missing `vocab_size` and `special_tokens` being a list instead of a dict are unrelated." }, { - "left": "issue:44355", - "right": "issue:44855", + "left": "issue:29870", + "right": "issue:36296", "accept": false, - "reason": "Running compiled Python files and a torch.jit.script parsing IndentationError are both import-time issues, but not the same underlying bug." + "reason": "Bart generation masking error and tensor-parallel training state issues are different subsystems." }, { - "left": "issue:29870", - "right": "issue:36296", + "left": "issue:44355", + "right": "issue:44855", "accept": false, - "reason": "Bart generate UnboundLocalError and a tensor-parallel training bug are unrelated model/runtime failures." + "reason": "Compiled-file execution errors and a Python 3.13 `@torch.jit.script` parsing/indentation issue are distinct." }, { "left": "issue:43867", "right": "issue:46202", "accept": false, - "reason": "Sorted state_dict loading and Gemma pipeline roundtrip on \u2581 are different serialization/tokenization issues." + "reason": "A sorted state-dict loading bug is not the same as Gemma pipeline roundtrip failure on the `\u2581` token." }, { "left": "issue:44908", "right": "issue:46851", "accept": false, - "reason": "Scheduler kwargs being ignored is unrelated to Qwen3.5 variable-length position_ids handling." + "reason": "Scheduler kwarg handling and Qwen3.5 variable-length `position_ids` behavior are unrelated." }, { "left": "issue:43992", "right": "issue:47317", "accept": false, - "reason": "UMT5Encoder missing embed_tokens.weight and PPFormulaNet training-loss bug are unrelated model-specific issues." + "reason": "Missing `embed_tokens.weight` in UMT5Encoder has nothing to do with PPFormulaNet training loss." }, { "left": "issue:44393", "right": "issue:47337", "accept": false, - "reason": "Qwen3-VL bbox hallucination/error and a flaky bbox-range test are not the same failure." - }, - { - "left": "issue:44075", - "right": "issue:46365", - "accept": false, - "reason": "SGD args not being used and Trainer import failing due to datasets are separate problems." - }, - { - "left": "issue:43010", - "right": "issue:45278", - "accept": false, - "reason": "A no_grad decoration issue for cache/layer updates is unrelated to broad import errors after upgrade." - }, - { - "left": "issue:44393", - "right": "issue:47000", - "accept": false, - "reason": "Qwen3-VL bbox output issues and RTDetrLoss infeasible cost matrix are different vision-model failures." - }, - { - "left": "issue:43992", - "right": "issue:45310", - "accept": false, - "reason": "UMT5Encoder missing embed_tokens.weight and Qwen3.5 MoE from_pretrained errors are different model-loading bugs." - }, - { - "left": "issue:42548", - "right": "issue:44351", - "accept": false, - "reason": "Both are import errors, but missing PreTrainedModel and missing HybridCache are distinct exported-symbol problems." - }, - { - "left": "issue:29870", - "right": "issue:33453", - "accept": false, - "reason": "Bart generate decoder_attention_mask bug and SigLIP tokenizer loading regression are unrelated." + "reason": "Qwen3-VL 2D bbox output issues and a flaky bbox-range test are not the same bug." } ] }, @@ -19211,60 +19088,66 @@ "cluster_id": "cluster-10000-540", "nodes": [ "issue:28282", + "issue:29870", "issue:33453", "issue:36296", "issue:41628", "issue:41669", "issue:42548", "issue:42890", - "issue:42898", "issue:43010", "issue:43064", "issue:43065", "issue:43502", "issue:43525", "issue:43575", - "issue:43576", "issue:43618", "issue:43723", - "issue:43824", "issue:43867", "issue:43976", + "issue:43992", "issue:44062", + "issue:44075", "issue:44077", "issue:44112", "issue:44222", "issue:44230", "issue:44246", + "issue:44351", "issue:44360", "issue:44387", - "issue:44485", + "issue:44393", "issue:44492", "issue:44556", "issue:44734", "issue:44964", "issue:44977", - "issue:45071", "issue:45125", + "issue:45278", + "issue:45310", "issue:45357", "issue:45375", "issue:45440", "issue:45478", "issue:45561", "issue:45850", - "issue:46132", - "issue:46143", - "issue:46144", + "issue:46365", "issue:46617", - "issue:46664", + "issue:47000", "issue:47317" ], "soft_pairs": [ + "issue:44075|issue:46365", + "issue:43010|issue:45278", + "issue:44393|issue:47000", + "issue:42548|issue:44351", + "issue:43992|issue:45310", + "issue:29870|issue:33453", "issue:43064|issue:44492", "issue:36296|issue:43575", "issue:44222|issue:45125", - "issue:41669|issue:44246", "issue:43618|issue:44077", + "issue:41669|issue:44246", "issue:42890|issue:45561", "issue:43867|issue:45357", "issue:43867|issue:45375", @@ -19284,234 +19167,234 @@ "issue:41628|issue:42548", "issue:43723|issue:44556", "issue:44222|issue:44230", - "issue:43010|issue:43065", - "issue:43824|issue:46617", - "issue:43576|issue:46132", - "issue:42898|issue:46664", - "issue:45071|issue:47317", - "issue:46143|issue:46144", - "issue:43502|issue:44485" + "issue:43010|issue:43065" ], - "prepared_review_unit_hash": "190859b9b467e3eef3aa00095961d3f389214fa95349ae93abc24856303fe3cf", + "prepared_review_unit_hash": "664ae10f597cdda87825de472784c84afb36186f0ce020128a82d0910c084a98", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36367, - "estimated_input_tokens": 9092, - "estimated_eval_tokens": 18440 + "serialized_chars": 36425, + "estimated_input_tokens": 9107, + "estimated_eval_tokens": 18470 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly heterogeneous issue reports across unrelated regressions. The only pair that looks like the same concrete bug is the Qwen3.5 MoE tensor-parallel load/from_pretrained failure; the rest are too broad or touch different code paths.", - "confidence": 0.63, - "canonical_issue_reason": "Issue 45125 is the most concrete root-cause report in the only plausible duplicate pair: it names the missing `_tp_plan` tensor-parallelism implementation behind the broader Qwen3.5 MoE from_pretrained failure.", + "summary": "This cluster is mostly a grab bag of distinct bugs, not one duplicate family. The only clear duplicate-like pair is the import-time regression reports in `issue:41669` and `issue:44246`; most other soft edges are too broad or point to different code paths, models, or failure modes.", + "confidence": 0.87, + "canonical_issue_reason": "`issue:41669` is the most concrete and root-cause-oriented issue in the cluster: it identifies a specific import-time regression (`import *` in model modules) and proposes a precise fix, which also aligns with the nearby slow-import report in `issue:44246`.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45125 is the best representative issue here because it is specific, diagnostic, and likely the actionable root cause for the related Qwen3.5 MoE loading error.", + "best_issue_reason": "`issue:41669` is the best representative issue because it is specific, actionable, and anchored to a clear code-path regression rather than a broad symptom report.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44075", + "right": "issue:46365", + "accept": false, + "reason": "Different problems: optimizer argument handling vs an import failure involving `Trainer` and `datasets`." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "Both mention regressions, but one is about cache update methods and the other is broad import errors after upgrade; not the same bug." + }, + { + "left": "issue:44393", + "right": "issue:47000", + "accept": false, + "reason": "Different model/task failures: Qwen3-VL bounding-box output issue vs RTDetr loss infeasibility." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and likely different root causes." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Different `from_pretrained` failures on different models; no evidence they share one concrete code-path bug." + }, + { + "left": "issue:29870", + "right": "issue:33453", + "accept": false, + "reason": "Bart generation UnboundLocalError and tokenizer-loading regression are unrelated." + }, { "left": "issue:43064", "right": "issue:44492", "accept": false, - "reason": "Different problems: FSDP/PEFT optimizer-state corruption vs a cache-strategy typo." + "reason": "Unrelated: FSDP/PEFT optimizer-state bug versus a cache-strategy typo." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Both mention tensor parallelism, but one is a training bug and the other is a model-loading OOM; not the same bug." + "reason": "Both involve tensor parallelism, but one is a training bug and the other is an OOM on loading a specific large model." }, { "left": "issue:44222", "right": "issue:45125", "accept": false, - "reason": "FP8 save_pretrained MoE bug vs tensor-parallel plan missing for Qwen3.5 MoE; different failure modes." - }, - { - "left": "issue:41669", - "right": "issue:44246", - "accept": false, - "reason": "Import-time slowdown and import-* cleanup are related broadly, but not clearly the same underlying bug/change." + "reason": "Different issues: FP8 `save_pretrained` behavior versus missing `_tp_plan` for tensor parallelism." }, { "left": "issue:43618", "right": "issue:44077", "accept": false, - "reason": "Unrelated model behaviors: CLIPOutput attentions vs PatchTSMixer post_init policy." + "reason": "Completely different model/output regression versus an API restriction on `patchtsmixer` post-init." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": true, + "reason": "Both describe the same underlying regression: `import transformers` becoming slow, with `issue:41669` pointing to a specific root cause in model imports." }, { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "Different test failures: seeding-related instability vs xdist file-race in testing utils." + "reason": "Different test failures: missing `set_seed()` in a model integration test vs xdist race on captured info output." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "Load error with sorted state_dict vs incorrect visual encoder keys on save_pretrained; not the same issue." + "reason": "Both are model loading/saving problems, but they affect different mechanisms and are not the same code-path." }, { "left": "issue:43867", "right": "issue:45375", "accept": false, - "reason": "State-dict sorting load error is unrelated to a missing vision config field." + "reason": "A generic state-dict load error is not the same as a missing vision-config field silently dropped by strict parsing." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "PyTorch missing ImportError vs tokenizer-loading regression; unrelated." + "reason": "PyTorch-not-found import error and tokenizer-loading regression are unrelated." }, { "left": "issue:43010", "right": "issue:43976", "accept": false, - "reason": "Torch-no_grad cache update change vs Python version compatibility issue; different code paths." + "reason": "Cache-update no-grad concern is unrelated to a Python version compatibility issue." }, { "left": "issue:43618", "right": "issue:44977", "accept": false, - "reason": "CLIP output attentions bug vs Qwen3.5 flash-attention generation issue; unrelated." + "reason": "Different failures in different models; no shared concrete bug." }, { "left": "issue:43010", "right": "issue:45850", "accept": false, - "reason": "A specific cache update change is not the same as the broad 'many models break' report." + "reason": "One is a specific cache-update bug; the other is a vague broad complaint about many models breaking." }, { "left": "issue:45125", "right": "issue:45478", - "accept": true, - "reason": "Same concrete Qwen3.5 MoE loading failure: the broad from_pretrained error is plausibly caused by the missing `_tp_plan` tensor-parallel implementation." + "accept": false, + "reason": "Related model family, but one is a missing tensor-parallel plan and the other is a generic `from_pretrained` error; too little to treat as the same bug." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Tensor-parallel `_tp_plan` omission vs missing `deepstack_visual_indexes` in vision config; different defects." + "reason": "Tensor-parallel plan missing and vision-config field dropping are different issues." }, { "left": "issue:44360", "right": "issue:45440", "accept": false, - "reason": "DSA indexer ReLU complaint is unrelated to DeepSeekV3MoE divergence." + "reason": "Different domains entirely: DSA indexer implementation vs DeepSeekV3MoE parity with remote code." }, { "left": "issue:43502", "right": "issue:44062", "accept": false, - "reason": "Local-files-only network leak vs AddedToken argument conflict; unrelated." + "reason": "Offline-download behavior and AddedToken constructor argument collision are unrelated." }, { "left": "issue:42890", "right": "issue:44112", "accept": false, - "reason": "Different CI/test issues with no shared code path." + "reason": "Separate flaky tests in different models and subsystems." }, { "left": "issue:43525", "right": "issue:47317", "accept": false, - "reason": "Llama4 config pad_token_id attribute error vs PPFormulaNet training-loss bug; unrelated." + "reason": "Llama4 config attribute error and PPFormulaNet training-loss bug do not share a code path." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "SamHQ test seeding issue vs Phi-4-multimodal model load failure; unrelated." + "reason": "Different model-specific test flake vs model loading failure." }, { "left": "issue:44734", "right": "issue:46617", "accept": false, - "reason": "KV-cache indexing crash in serve vs trust_remote_code cache FileNotFoundError; different bugs." + "reason": "KV-cache continuation crash and cached trust-remote-code file lookup failure are unrelated." }, { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Int4 quantization memory regression vs checkpoint reload compatibility after upgrade; different symptoms and likely causes." + "reason": "Different regressions: int4 quantization memory growth vs reloading older checkpoints in v5." }, { "left": "issue:41628", "right": "issue:42548", "accept": false, - "reason": "Both are import errors from transformers, but for different symbols and likely different missing exports." + "reason": "Both are import errors, but they concern different exported symbols and likely different breakages." }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Tokenizer loading regression vs checkpoint reload failure after upgrade; too broad to call duplicates." + "reason": "Tokenizer-loading issue and checkpoint reload incompatibility are distinct failures." }, { "left": "issue:44222", "right": "issue:44230", "accept": false, - "reason": "FP8 save_pretrained bug vs FP8 inference support request; not the same underlying change." - }, - { - "left": "issue:43010", - "right": "issue:43065", - "accept": false, - "reason": "Cache update no_grad behavior vs a dummy Conv2d in Sam3PixelDecoder; unrelated." - }, - { - "left": "issue:43824", - "right": "issue:46617", - "accept": false, - "reason": "Missing Qwen2_5_VLForConditionalGeneration import vs remote-code cache FileNotFoundError; different issues." - }, - { - "left": "issue:43576", - "right": "issue:46132", - "accept": false, - "reason": "Broken env command and AttentionInterface.register behavior are unrelated." + "reason": "FP8 save behavior and FP8 inference support are not the same bug." }, { - "left": "issue:42898", - "right": "issue:46664", + "left": "issue:42890", + "right": "issue:44112", "accept": false, - "reason": "Tokenization cleanup behavior change vs SmolVLM num_frames semantics; unrelated." + "reason": "Duplicate of an earlier reject: unrelated test failures." }, { - "left": "issue:45071", + "left": "issue:43502", "right": "issue:47317", "accept": false, - "reason": "PretrainedConfig type-checking regression vs PPFormulaNet training-loss bug; unrelated." - }, - { - "left": "issue:46143", - "right": "issue:46144", - "accept": false, - "reason": "Both are RoFormer-related, but one is about kwargs forwarding and the other about attention interface usage; not clearly the same bug." - }, - { - "left": "issue:43502", - "right": "issue:44485", - "accept": false, - "reason": "local_files_only network leak vs GLM-5 RoPE implementation discussion; unrelated." + "reason": "Offline-loading bug and PPFormulaNet training-loss bug are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded overall. The verdicts stay conservative, and the only accepted soft pair is defensible from the packet: issue 45125 is a specific root-cause report for the Qwen3.5 MoE from_pretrained failure seen in issue 45478. The remaining pairs are reasonably rejected as different bugs or code paths." + "feedback": "The clustering summary is grounded in the packet and stays conservative: it treats most pairs as distinct issues unless the titles/body clearly align. The lone accepted edge, `issue:41669`/`issue:44246`, is reasonably supported by the shared import-slowdown symptom and the specific import-* root cause in 41669. No soft-edge verdict overstates evidence." }, "evaluator_used": true, "retried": false, @@ -19539,15 +19422,13 @@ "issue:36296", "issue:38617", "issue:42757", + "issue:42898", "issue:43010", "issue:43232", "issue:43299", - "issue:43352", "issue:43475", "issue:43502", - "issue:43525", "issue:43576", - "issue:43673", "issue:43824", "issue:43827", "issue:43874", @@ -19561,7 +19442,6 @@ "issue:44355", "issue:44360", "issue:44485", - "issue:44492", "issue:44556", "issue:44623", "issue:44749", @@ -19574,16 +19454,26 @@ "issue:45375", "issue:45850", "issue:46095", + "issue:46132", + "issue:46143", + "issue:46144", "issue:46617", "issue:46664", "issue:46726", "issue:47059", - "issue:47110" + "issue:47110", + "issue:47317" ], "soft_pairs": [ + "issue:43824|issue:46617", + "issue:43576|issue:46132", + "issue:45071|issue:47317", + "issue:42898|issue:46664", + "issue:46143|issue:46144", + "issue:43502|issue:44485", "issue:44355|issue:44749", - "issue:30990|issue:31795", "issue:43874|issue:46617", + "issue:30990|issue:31795", "issue:45850|issue:47110", "issue:43010|issue:43232", "issue:43502|issue:44222", @@ -19609,231 +19499,246 @@ "issue:44315|issue:44360", "issue:45362|issue:45850", "issue:45375|issue:47110", - "issue:43673|issue:44492", - "issue:44485|issue:44961", - "issue:43352|issue:44961", - "issue:43525|issue:47110" + "issue:44485|issue:44961" ], - "prepared_review_unit_hash": "fc0ebd1cf4143e760128128ba3a76ecb50f0b477eac7737f5a42aa3b6292f1fb", + "prepared_review_unit_hash": "a0e5c8d3e568b820b2920f7bf773bc04a7f02c77f520bda4c2dc8c2d299d6083", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 32, - "serialized_chars": 36278, - "estimated_input_tokens": 9070, - "estimated_eval_tokens": 18396 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37721, + "estimated_input_tokens": 9431, + "estimated_eval_tokens": 19118 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated Transformer issues. None of the soft pairs are strong duplicates from titles alone; the clearest representative issue is the local_files_only regression because it states a precise contract violation with an obvious reproduction path.", - "confidence": 0.86, - "canonical_issue_reason": "issue:43502 is the most concrete and actionable bug report here: it describes a specific regression where `local_files_only=True` still triggers API requests.", + "summary": "This cluster is dominated by loose textual similarity around Transformers v5 regressions, but the suggested pairs do not describe the same underlying bug/change. I would not merge any of the soft edges.", + "confidence": 0.95, + "canonical_issue_reason": "Issue 43576 is the best representative of the loose cluster because it is a concrete v5 regression and acts as a hub in the similarity graph, but it is still not a true duplicate anchor.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43502 is the strongest canonical-style issue because it is narrowly scoped, user-facing, and easy to validate; the other issues are either model-specific, broader compatibility reports, or documentation problems.", + "best_issue_reason": "43576 is the clearest, most concrete issue here; the rest are mostly unrelated breakages, docs problems, or different model-specific failures.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44355", - "right": "issue:44749", + "left": "issue:43824", + "right": "issue:46617", "accept": false, - "reason": "Different problems: compiled Python file execution errors vs a performance regression after upgrading Transformers." + "reason": "Different failures: one is a missing Qwen import, the other is a cache FileNotFoundError for trust_remote_code." }, { - "left": "issue:30990", - "right": "issue:31795", + "left": "issue:43576", + "right": "issue:46132", + "accept": false, + "reason": "One is a broken CLI/env command in v5; the other is an AttentionInterface.register behavior change." + }, + { + "left": "issue:45071", + "right": "issue:47317", + "accept": false, + "reason": "PretrainedConfig type-checking and PPFormulaNet training loss are unrelated problems." + }, + { + "left": "issue:42898", + "right": "issue:46664", + "accept": false, + "reason": "Tokenization cleanup behavior and SmolVLM video frame semantics are different subsystems and bugs." + }, + { + "left": "issue:46143", + "right": "issue:46144", + "accept": false, + "reason": "Both involve RoFormer, but one is kwargs propagation and the other is attention-interface usage; not the same bug." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "local_files_only network leakage and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44749", "accept": false, - "reason": "One is a loading hang in Sentence Transformers; the other is documentation confusion about `input_ids` and `past_key_values`." + "reason": "Compiled Python file errors and a performance regression after upgrade are not the same issue." }, { "left": "issue:43874", "right": "issue:46617", "accept": false, - "reason": "Different code paths: missing image-processor method vs FileNotFoundError loading a `trust_remote_code` model from cache." + "reason": "A missing image-processor method is unrelated to failing to load a trust_remote_code model from cache." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Loading hang for Sentence Transformers and documentation confusion in model.forward are different topics." }, { "left": "issue:45850", "right": "issue:47110", "accept": false, - "reason": "Too broad and unrelated: a general v5 breakage complaint vs a `special_tokens` attribute error." + "reason": "Broad v5 complaints do not match a specific special_tokens AttributeError." }, { "left": "issue:43010", "right": "issue:43232", "accept": false, - "reason": "Both involve generation internals, but one is about `@torch.no_grad` on cache/layer updates and the other is a sync_gpus generation bug." + "reason": "Both touch generation internals, but one is about no_grad decoration and the other about sync_gpus model-kwargs updates." }, { "left": "issue:43502", "right": "issue:44222", "accept": false, - "reason": "Local-files-only network leakage is unrelated to FP8 `save_pretrained` behavior for MoE models." + "reason": "local_files_only API requests and FP8 save_pretrained for moe are unrelated." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Both mention v5 upgrade regressions, but one is a Qwen3-VL-Moe load failure and the other is a general checkpoint reload problem." + "reason": "Qwen3-VL-Moe loading and reloading old checkpoints after upgrade are different concrete failures." }, { "left": "issue:46664", "right": "issue:46726", "accept": false, - "reason": "SmolVLM `num_frames` semantics are unrelated to AutoProcessor backend-missing error handling." + "reason": "SmolVLM video-frame semantics and missing vision backend handling are distinct bugs." }, { "left": "issue:30333", "right": "issue:36296", "accept": false, - "reason": "MLflow reporting on failed jobs is unrelated to a tensor-parallel training bug." + "reason": "MLflow reporting on training failure and tensor-parallel training bugs are unrelated." }, { "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "Documentation references to `pipeline()` have nothing to do with the unrelated `racoon` report." + "reason": "Docs still mentioning pipeline() and the 'racoon' issue are not the same change or bug." }, { "left": "issue:44230", "right": "issue:46095", "accept": false, - "reason": "FP8 inference support for Qwen models is not the same underlying issue as a DeepSeekV4 weight-conversion question." + "reason": "FP8 Qwen3-VL/moe support and DeepSeekV4 weight conversion are different requests." }, { "left": "issue:29870", "right": "issue:31795", "accept": false, - "reason": "A runtime `decoder_attention_mask` error is not the same as documentation confusion about generation inputs." + "reason": "A generation UnboundLocalError in Bart and forward-argument documentation confusion are unrelated." }, { "left": "issue:44038", "right": "issue:45125", "accept": false, - "reason": "Both are Qwen-related, but one is a broad transformers-v5 bug report and the other is a missing `_tp_plan` tensor-parallel implementation detail." + "reason": "Both involve Qwen3.5/Moe wording, but one is a loading bug and the other is missing tensor-parallel plan support." }, { "left": "issue:43576", "right": "issue:43824", "accept": false, - "reason": "A broken `transformers env` command is not the same failure as an import error for a specific model class." + "reason": "CLI/env breakage and a specific import error for Qwen2.5-VL are different failures." }, { "left": "issue:43502", "right": "issue:44961", "accept": false, - "reason": "`local_files_only` network leakage and an unrelated placeholder issue are clearly different." + "reason": "local_files_only behavior and an unrelated 'racoon' issue do not match." }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Compiled-file execution errors are unrelated to `PretrainedConfig` type-checking regressions." + "reason": "Compiled-file runtime errors and PretrainedConfig type checking are unrelated regressions." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "A CLI environment command failure is not the same as a type-checking breakage in `PretrainedConfig`." + "reason": "A broken env command is not the same as type-checking failures in PretrainedConfig." }, { "left": "issue:43576", "right": "issue:47059", "accept": false, - "reason": "The `transformers env` command issue is different from the `huggingface_hub` CLI completion incompatibility." + "reason": "CLI/env breakage and a huggingface_hub completion attribute mismatch are separate issues." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "An `is_offline_mode` import error may affect CLI behavior, but it is not the same concrete bug as the v5 env-command report." + "reason": "Missing is_offline_mode import and broken transformers env are different problems." }, { "left": "issue:36296", "right": "issue:43901", "accept": false, - "reason": "Tensor-parallel training issues are unrelated to `TextClassificationPipeline` docs still mentioning `return_all_scores`." + "reason": "Tensor-parallel training and text-classification docs mentioning return_all_scores are unrelated." }, { "left": "issue:31795", "right": "issue:38617", "accept": false, - "reason": "Documentation confusion about generation args is unrelated to a missing `layer_type_validation` import." + "reason": "A docs clarification issue and an ImportError for layer_type_validation are not duplicates." }, { "left": "issue:43475", "right": "issue:46617", "accept": false, - "reason": "SAM3 video attribute loss is unrelated to a cache FileNotFoundError for trust-remote-code models." + "reason": "SAM 3 Video attribute loss and trust_remote_code cache loading failure are unrelated." }, { "left": "issue:44077", "right": "issue:44861", "accept": false, - "reason": "A `post_init` allowance policy issue is unrelated to `_get_tied_weight_keys` crashing on a list." + "reason": "A patchtsmixer post_init policy question and a tied-weights AttributeError are different bugs." }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "Missing processor save files and a generic bug report are not the same underlying issue." + "reason": "Missing processor.save_pretrained files and a generic bug report are not the same underlying issue." }, { "left": "issue:44351", "right": "issue:46617", "accept": false, - "reason": "HybridCache import failure and cache FileNotFoundError are different cache-loading problems." + "reason": "HybridCache import failure and cache FileNotFoundError are distinct failures." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Liger Kernel application in `model_init` is unrelated to a DSA indexer/ReLU discussion." + "reason": "Liger Kernel application with model_init and a DSA indexer ReLU discussion are unrelated." }, { "left": "issue:45362", "right": "issue:45850", "accept": false, - "reason": "Qwen3.5 chat crashes are model-specific, while the other issue is a broad complaint that v5 breaks many models." + "reason": "A Qwen3.5-35B chat crash is not the same as a broad complaint that v5 breaks many models." }, { "left": "issue:45375", "right": "issue:47110", "accept": false, - "reason": "A missing config field in Qwen3.5 Moe vision is unrelated to a `special_tokens` list/type error." - }, - { - "left": "issue:43673", - "right": "issue:44492", - "accept": false, - "reason": "A missing GenerationMixin cache during chunked prefill is not the same as a typo in cache strategy docs/code." + "reason": "Missing Qwen3.5MoeVisionConfig field and special_tokens list-key access error are different bugs." }, { "left": "issue:44485", "right": "issue:44961", "accept": false, - "reason": "GLM-5 RoPE implementation discussion is unrelated to the placeholder `racoon` issue." - }, - { - "left": "issue:43352", - "right": "issue:44961", - "accept": false, - "reason": "Flash-Attention support for Nemotron is unrelated to the placeholder `racoon` issue." - }, - { - "left": "issue:43525", - "right": "issue:47110", - "accept": false, - "reason": "A missing `pad_token_id` on `Llama4Config` is not the same as a `special_tokens` list attribute error." + "reason": "GLM-5 RoPE implementation and the 'racoon' issue do not match." } ] }, @@ -19861,6 +19766,7 @@ "issue:42371", "issue:42890", "issue:43298", + "issue:43352", "issue:43408", "issue:43525", "issue:43572", @@ -19892,7 +19798,6 @@ "issue:45381", "issue:45468", "issue:45542", - "issue:45656", "issue:45753", "issue:45850", "issue:46018", @@ -19907,12 +19812,15 @@ "issue:47176" ], "soft_pairs": [ + "issue:43673|issue:44492", + "issue:43352|issue:44961", + "issue:43525|issue:47110", "issue:44273|issue:45542", "issue:44961|issue:45468", - "issue:45381|issue:46664", "issue:44315|issue:46209", - "issue:43867|issue:47110", + "issue:45381|issue:46664", "issue:44273|issue:45003", + "issue:43867|issue:47110", "issue:44466|issue:44556", "issue:41669|issue:43673", "issue:44655|issue:44861", @@ -19920,10 +19828,10 @@ "issue:43525|issue:45850", "issue:44485|issue:46664", "issue:44393|issue:45753", - "issue:44295|issue:45230", "issue:44393|issue:45071", - "issue:44297|issue:45230", + "issue:44295|issue:45230", "issue:44556|issue:45092", + "issue:44297|issue:45230", "issue:42371|issue:44492", "issue:45850|issue:47059", "issue:44246|issue:44273", @@ -19931,264 +19839,268 @@ "issue:43408|issue:44315", "issue:44492|issue:45850", "issue:41669|issue:46218", - "issue:44492|issue:46328", "issue:43976|issue:44393", + "issue:44492|issue:46328", "issue:42890|issue:44908", "issue:44704|issue:44908", - "issue:41669|issue:45341", "issue:44908|issue:45071", - "issue:43298|issue:46095", + "issue:41669|issue:45341", "issue:41669|issue:44351", + "issue:43298|issue:46095", "issue:43576|issue:44908", "issue:43576|issue:47176", - "issue:43976|issue:46218", - "issue:45230|issue:45656", - "issue:43976|issue:47176" + "issue:43976|issue:46218" ], - "prepared_review_unit_hash": "ca2b5042c214a53c7a69e37da76b5f8995cbae976d31760e5389dede59a3077e", + "prepared_review_unit_hash": "93366884b147c05427d5686c215bcfefc1b3a7d6eda0b619ae176d3877e85632", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 37, - "serialized_chars": 38126, - "estimated_input_tokens": 9532, - "estimated_eval_tokens": 19320 + "soft_pair_count": 38, + "serialized_chars": 38445, + "estimated_input_tokens": 9612, + "estimated_eval_tokens": 19480 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of Transformers v5 regression/compatibility issues, but the individual reports cover different models, APIs, and failure modes rather than one exact bug.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 45850 is the broadest umbrella for the recurring v5 breakage theme and can represent the cluster better than the more narrowly scoped reports.", + "summary": "This cluster is a grab-bag of mostly unrelated Transformers issues, many around v5 regressions and model-specific bugs. None of the soft pairs look like clear same-bug duplicates, so I reject all candidate edges.", + "confidence": 0.39, + "canonical_issue_reason": "No single duplicate hub stands out; issue 44556 is the broadest, most user-facing v5 checkpoint reload regression in the set and is the best representative issue.", "canonical_pr_reason": null, - "best_issue_reason": "45850 is the most representative high-level issue for the shared upgrade/regression pattern, even though it is vague.", + "best_issue_reason": "Issue 44556 is the strongest representative because it captures a broad, concrete v5 compatibility failure around reloading checkpoints, unlike the more model-specific or narrowly scoped reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Chunked-prefill cache missing vs a cache-strategy typo; same vocabulary, different bug." + }, + { + "left": "issue:43352", + "right": "issue:44961", + "accept": false, + "reason": "Flash Attention support for Nemotron vs an unrelated placeholder issue title." + }, + { + "left": "issue:43525", + "right": "issue:47110", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config vs special_tokens dict/list handling; different code paths." + }, { "left": "issue:44273", "right": "issue:45542", "accept": false, - "reason": "Different bugs: lazy loading behavior vs TensorFlow backend detection/install handling." + "reason": "Lazy loading regression vs TensorFlow backend detection error; unrelated symptoms and fixes." }, { "left": "issue:44961", "right": "issue:45468", "accept": false, - "reason": "Unrelated; one is a placeholder issue, the other is a Gemma-4 audio positional encoding bug." - }, - { - "left": "issue:45381", - "right": "issue:46664", - "accept": false, - "reason": "Both involve multimodal/video handling, but they target different models and concrete semantics." + "reason": "Placeholder issue vs Gemma-4 audio positional encoding bug; no shared underlying defect." }, { "left": "issue:44315", "right": "issue:46209", "accept": false, - "reason": "Different code paths: model_init application of Liger Kernel vs FP8 inference on B200." + "reason": "Liger Kernel/model_init wiring vs FP8 inference on B200; different subsystems and failures." }, { - "left": "issue:43867", - "right": "issue:47110", + "left": "issue:45381", + "right": "issue:46664", "accept": false, - "reason": "Load/state_dict sorting error is unrelated to special_tokens dict/list handling." + "reason": "Qwen2.5-VL video position IDs vs SmolVLM num_frames semantics; both multimodal, but not the same bug." }, { "left": "issue:44273", "right": "issue:45003", "accept": false, - "reason": "Import performance/lazy-loading is not the same bug as unsafe sys.modules access." + "reason": "Lazy loading/import behavior vs unsafe sys.modules access; different root causes." + }, + { + "left": "issue:43867", + "right": "issue:47110", + "accept": false, + "reason": "State-dict loading error vs special_tokens structure error; unrelated load/save paths." }, { "left": "issue:44466", "right": "issue:44556", "accept": false, - "reason": "Both touch v5 compatibility, but one is tied-weight serialization and the other is checkpoint reload compatibility." + "reason": "Serialization inconsistency in tied weights vs generic checkpoint reload incompatibility; related area, but not the same concrete bug." }, { "left": "issue:41669", "right": "issue:43673", "accept": false, - "reason": "Import-star cleanup/perf issue is unrelated to chunked_prefill cache availability." + "reason": "Import-* cleanup/performance issue vs missing generation cache in chunked_prefill; different problems." }, { "left": "issue:44655", "right": "issue:44861", "accept": false, - "reason": "Saving Pipeline objects and tied-weight key extraction are different failures." + "reason": "Pipeline save_pretrained failure vs _tied_weights_keys list crash; different save-time code paths." }, { "left": "issue:43572", "right": "issue:45850", "accept": false, - "reason": "Specific missing config attribute vs a broad complaint about many v5 regressions." + "reason": "StableLmConfig pad_token_idx regression vs broad v5 breakage claim; too general and not the same bug." }, { "left": "issue:43525", "right": "issue:45850", "accept": false, - "reason": "Specific Llama4Config attribute error, not the same underlying bug as the generic v5 breakage report." + "reason": "Llama4Config missing pad_token_id vs broad v5 breakage report; not a duplicate." }, { "left": "issue:44485", "right": "issue:46664", "accept": false, - "reason": "Different model implementations and different bugs (RoPE vs video frame semantics)." + "reason": "GLM-5 RoPE implementation discussion vs SmolVLM video-frame semantics; unrelated." }, { "left": "issue:44393", "right": "issue:45753", "accept": false, - "reason": "Same family (Qwen vision) but different models and failure modes; not one concrete bug." - }, - { - "left": "issue:44295", - "right": "issue:45230", - "accept": false, - "reason": "A position_ids buffer access error is not the same as an unspecified bug report." + "reason": "2D bbox output issue in Qwen3-VL vs infinite loop on a specific image in Qwen3.5; different failure modes." }, { "left": "issue:44393", "right": "issue:45071", "accept": false, - "reason": "Different issues: Qwen3-VL output bug vs PretrainedConfig type checking regression." + "reason": "Model output bug vs PretrainedConfig type-checking regression; unrelated." }, { - "left": "issue:44297", + "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to the generic bug report." + "reason": "position_ids buffer access error vs generic 'Bug report' placeholder; no evidence of same bug." }, { - "left": "issue:44556", - "right": "issue:45092", + "left": "issue:44297", + "right": "issue:45230", "accept": false, - "reason": "Both are upgrade compatibility complaints, but they concern different model families and mechanics." + "reason": "tokenizer.save_pretrained mismatch vs generic report placeholder; unrelated." }, { "left": "issue:42371", "right": "issue:44492", "accept": false, - "reason": "TF32 settings API guidance is unrelated to a typo in cache strategies." + "reason": "TF32 configuration API change vs cache strategy typo; different areas entirely." }, { "left": "issue:45850", "right": "issue:47059", "accept": false, - "reason": "A broad v5 regression report does not match a specific CLI break from huggingface_hub 1.22.0." + "reason": "Broad v5 regression complaint vs cli failure with huggingface_hub; no shared code-path." }, { "left": "issue:44246", "right": "issue:44273", "accept": false, - "reason": "Import slowness and lazy-loading malfunction are related only at a high level, not the same concrete bug." + "reason": "Import slowdown sometimes vs lazy loading not working properly; possibly adjacent, but not clearly the same bug." }, { "left": "issue:46018", "right": "issue:46172", "accept": false, - "reason": "Different DeepSeek-V4 problems: gating logic vs tensor dtype/transpose operation." + "reason": "DeepSeek-V4 shared expert gating vs a tensor transpose issue; unrelated implementation bugs." }, { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "Completely different subsystems and failures; no shared code-path bug." + "reason": "sam3_video/sam3_tracker type warning vs Liger Kernel not applied with model_init; different components." }, { "left": "issue:44492", "right": "issue:45850", "accept": false, - "reason": "A typo in cache strategies is not the same as the broad v5 regression umbrella." + "reason": "Cache-strategy typo vs broad v5 breakage report; not the same underlying defect." }, { "left": "issue:41669", "right": "issue:46218", "accept": false, - "reason": "Import-star usage/performance issue is unrelated to a missing exported model class." + "reason": "Import-* performance cleanup vs module import failure for Qwen3_5ForCausalLM; unrelated." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python version compatibility and a Qwen3-VL vision bug are different issues." + "reason": "Python version compatibility issue vs Qwen3-VL bounding-box hallucination; unrelated." }, { - "left": "issue:42890", - "right": "issue:44908", + "left": "issue:44492", + "right": "issue:46328", "accept": false, - "reason": "Test seeding flakiness and scheduler kwarg propagation are unrelated." + "reason": "Typo in cache strategies vs SequenceFeatureExtractor.pad numpy/list performance issue; different code paths." }, { - "left": "issue:44704", + "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "AutoProcessor kwargs forwarding is unrelated to inverse_sqrt scheduler behavior." + "reason": "Missing set_seed in tests vs scheduler kwargs being ignored; unrelated test vs scheduler bug." }, { - "left": "issue:41669", - "right": "issue:45341", + "left": "issue:44704", + "right": "issue:44908", "accept": false, - "reason": "Import cleanup/performance issue is unrelated to a testing_utils bug." + "reason": "AutoProcessor kwargs not forwarded to cached_file vs inverse_sqrt scheduler kwargs ignored; both kwargs issues, but different APIs." }, { "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwargs handling and PretrainedConfig type checking are different code paths." + "reason": "Scheduler kwargs bug vs PretrainedConfig type-checking regression; different subsystems." }, { - "left": "issue:43298", - "right": "issue:46095", + "left": "issue:41669", + "right": "issue:45341", "accept": false, - "reason": "Continuous batching audio support and DeepSeek-V4 weight conversion are unrelated." + "reason": "Import-* cleanup/performance issue vs a small testing_utils bug; not the same bug." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Import performance cleanup is not the same as failing to export HybridCache." + "reason": "Import-* cleanup/performance issue vs HybridCache import failure; unrelated." + }, + { + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Audio continuous batching support vs DeepSeek-V4 weight conversion question; different topics." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "Broken env command and scheduler kwargs propagation are different bugs." + "reason": "transformers env CLI breakage vs scheduler kwargs bug; unrelated." }, { "left": "issue:43576", "right": "issue:47176", "accept": false, - "reason": "CLI env command breakage is unrelated to a shell command-substitution allowlist bypass." + "reason": "Broken env command vs shell-operator allowlist bypass security bug; different command-line paths and different severity." }, { "left": "issue:43976", "right": "issue:46218", "accept": false, - "reason": "Python compatibility and missing model export are unrelated." - }, - { - "left": "issue:45230", - "right": "issue:45656", - "accept": false, - "reason": "Generic bug report vs optimizer step being called twice under deepspeed." - }, - { - "left": "issue:43976", - "right": "issue:47176", - "accept": false, - "reason": "Python support issue and shell allowlist bypass are unrelated." + "reason": "Python version compatibility regression vs module import error for Qwen3_5ForCausalLM; unrelated." } ] }, @@ -20214,11 +20126,9 @@ "nodes": [ "issue:28348", "issue:29870", - "issue:30333", "issue:31515", "issue:31795", "issue:33453", - "issue:36296", "issue:37428", "issue:38617", "issue:41669", @@ -20247,11 +20157,13 @@ "issue:44861", "issue:44908", "issue:44998", + "issue:45230", "issue:45250", "issue:45290", "issue:45310", "issue:45341", "issue:45468", + "issue:45656", "issue:45850", "issue:46218", "issue:46655", @@ -20262,6 +20174,8 @@ "issue:47177" ], "soft_pairs": [ + "issue:45230|issue:45656", + "issue:43976|issue:47176", "issue:31795|issue:33453", "issue:43519|issue:46655", "issue:44315|issue:44829", @@ -20306,327 +20220,336 @@ "issue:43618|issue:44998", "issue:31515|issue:38617", "issue:29870|issue:31515", - "issue:43298|issue:44230", - "issue:30333|issue:31515", - "issue:31515|issue:36296", - "issue:31515|issue:31795" + "issue:43298|issue:44230" ], - "prepared_review_unit_hash": "37e4775e78b8143655ab0fe3dc106669ccb2e54dcce50e5a80951a36c766727b", + "prepared_review_unit_hash": "2fd0aa8e595b50afd22519a9f6cc183f077103b13d6432da86739ec5ad927ac5", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 48, - "serialized_chars": 41345, - "estimated_input_tokens": 10337, - "estimated_eval_tokens": 20930 + "soft_pair_count": 47, + "serialized_chars": 40992, + "estimated_input_tokens": 10248, + "estimated_eval_tokens": 20752 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heavily over-broadened: most items are unrelated Transformers issues, with only a loose recurring theme around v5/model-loading/tokenizer regressions. No soft pair looks like a true duplicate with the same concrete code path.", - "confidence": 0.33, - "canonical_issue_reason": "issue:45850 is the broadest umbrella for the recurring v5 breakage theme, so it is the best representative issue even though it is very generic.", + "summary": "This cluster is a grab bag of mostly unrelated issues. There is some loose overlap around Flash Attention 2.0, tokenizer/model-loading regressions, and a few processor bugs, but none of the soft pairs look like the same concrete underlying bug or change. I would not merge any of the suggested pairs.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 45250 is the broadest and most central Flash Attention 2.0 request in the set, so it is the best umbrella candidate among otherwise unrelated issues.", "canonical_pr_reason": null, - "best_issue_reason": "Best overall issue to anchor the cluster because it captures the widest shared regression theme across the set.", + "best_issue_reason": "Issue 45250 is the most representative high-level issue here; it is generic enough to cover the closest related flash-attention requests without overfitting to a specific model.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45230", + "right": "issue:45656", + "accept": false, + "reason": "Generic bug-report placeholder versus a specific DeepSpeed optimizer-step bug; no shared code path." + }, + { + "left": "issue:43976", + "right": "issue:47176", + "accept": false, + "reason": "Python version incompatibility and shell allowlist bypass are unrelated problems." + }, { "left": "issue:31795", "right": "issue:33453", "accept": false, - "reason": "Documentation confusion vs a tokenizer-loading regression; different problem and fix surface." + "reason": "Docs confusion about forward inputs versus a tokenizer loading regression; different issues." }, { "left": "issue:43519", "right": "issue:46655", "accept": false, - "reason": "Both are processor bugs, but one is timestamp math and the other is padding math in a different path." + "reason": "Both involve processor calculations, but for different models and different bugs (timestamp vs padding)." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Different subsystems: Liger/model_init application versus flash-attention training degradation." + "reason": "Liger Kernel application in model_init is unrelated to flash_attention_3 training degeneration." }, { "left": "issue:44861", "right": "issue:47110", "accept": false, - "reason": "Separate AttributeErrors in unrelated helpers; not the same failing code path." + "reason": "Different AttributeErrors in different code paths; no evidence they are the same bug." }, { "left": "issue:45468", "right": "issue:46894", "accept": false, - "reason": "Different model families and different numerical bugs; no shared concrete change." + "reason": "Gemma audio positional encoding and CohereASR training loss shift are distinct model-specific bugs." }, { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "Both mention tokenizer v5 behavior, but the reported failures are different enough that a shared code path is not clear." + "reason": "Both are tokenizer v5-related, but one is a broad behavior difference and the other is a decoder-type bug; not clearly the same fix." }, { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "DSA indexer ReLU issue vs Gemma audio positional encoding bug; unrelated." + "reason": "DSA indexer ReLU discussion and Gemma audio positional encoding are unrelated." }, { "left": "issue:44485", "right": "issue:47176", "accept": false, - "reason": "RoPE implementation issue versus shell allowlist security bypass; unrelated." + "reason": "RoPE implementation and shell command-substitution bypass are entirely different areas." }, { "left": "issue:43976", "right": "issue:47177", "accept": false, - "reason": "Python version compatibility bug versus tar/zip slip RCE; unrelated." + "reason": "Python version support and tar/zip slip security issue do not match." }, { "left": "issue:43519", - "right": "issue:44485", + "right": "issue:44908", "accept": false, - "reason": "Processor timestamp bug versus RoPE implementation bug; different code paths." + "reason": "Processor timestamp math versus scheduler kwargs handling are unrelated." }, { "left": "issue:31795", "right": "issue:37428", "accept": false, - "reason": "Doc clarification issue versus import failure for a flash-attention helper; not duplicates." + "reason": "Documentation confusion and import error for flash support helper are not the same defect." }, { "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "Broken env command and lazy loading are different features, despite both surfacing in v5." + "reason": "Broken env command and lazy loading failure are separate behaviors." }, { "left": "issue:44485", "right": "issue:47177", "accept": false, - "reason": "Model positional encoding issue versus security file-write exploit; unrelated." + "reason": "RoPE implementation bug and archive path traversal are unrelated." }, { "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Lazy loading regression and scheduler kwargs handling are unrelated bugs." + "reason": "Lazy loading and inverse_sqrt scheduler kwargs are different subsystems." }, { "left": "issue:43519", "right": "issue:44908", "accept": false, - "reason": "Processor timestamp calculation versus scheduler parameter handling; unrelated." + "reason": "Processor timestamp calculation does not overlap with LR scheduler kwargs handling." }, { "left": "issue:44861", "right": "issue:47059", "accept": false, - "reason": "Different failures in different utilities; no shared underlying bug." + "reason": "Tied-weight key handling and CLI completion breakage are unrelated." }, { "left": "issue:41669", "right": "issue:43704", "accept": false, - "reason": "Import-star cleanup/perf issue versus a VRAM leak in multithreaded dataloading; unrelated." + "reason": "Import-star cleanup and VRAM leak in Qwen3ForCausalLM are not the same bug." }, { "left": "issue:43576", "right": "issue:46218", "accept": false, - "reason": "CLI env command breakage and model import failure are different regressions." + "reason": "Broken env command and Qwen3_5ForCausalLM import failure are distinct issues." }, { "left": "issue:44485", "right": "issue:45468", "accept": false, - "reason": "RoPE implementation and audio positional encoding are both positional topics, but not the same bug." + "reason": "Different model implementations and different failure modes." }, { "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "CPU offload device mismatch versus audio positional encoding; unrelated." + "reason": "CPU offload device mismatch is unrelated to Gemma audio positional encoding." }, { "left": "issue:42371", "right": "issue:43704", "accept": false, - "reason": "TF32 settings warning versus VRAM leak; different concerns." + "reason": "TF32 settings guidance and VRAM leak are unrelated." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "VRAM leak and RoPE implementation are unrelated." + "reason": "VRAM leak and RoPE implementation bug are not the same code-path." }, { "left": "issue:45850", "right": "issue:47177", "accept": false, - "reason": "Generic v5 breakage report versus a security file-write exploit; unrelated." + "reason": "General breakage report versus archive traversal security issue; no shared fix." }, { "left": "issue:45850", "right": "issue:47176", "accept": false, - "reason": "Generic model-breakage umbrella versus shell-command allowlist bypass; unrelated." + "reason": "Broad regression complaint and shell allowlist bypass are unrelated." }, { "left": "issue:43352", "right": "issue:47177", "accept": false, - "reason": "Flash-attention support gap versus arbitrary file-write exploit; unrelated." + "reason": "Flash Attention 2 support request for Nemotron and tar-slip security bug are unrelated." }, { "left": "issue:43502", "right": "issue:43519", "accept": false, - "reason": "Unexpected network access during local loading versus processor timestamp bug; unrelated." + "reason": "Unexpected network access and timestamp calculation are different bugs." }, { "left": "issue:43519", "right": "issue:44623", "accept": false, - "reason": "Timestamp calculation bug versus missing save_pretrained outputs; different behavior." + "reason": "Incorrect timestamp math and missing save_pretrained files are separate processor issues." }, { "left": "issue:28348", "right": "issue:45250", "accept": false, - "reason": "Both mention Flash Attention 2.0, but one is GPT-2 support and the other is an unrelated generic request." + "reason": "Both mention Flash Attention 2.0, but one is GPT2-specific support while the other is a generic title; not enough to prove the same concrete change." }, { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "Different numerical/configuration issues in different models; not the same bug." + "reason": "Different numerical precision/config issues in different model families." }, { "left": "issue:45850", "right": "issue:46218", "accept": false, - "reason": "Broad v5 regressions umbrella versus a specific model import failure; not the same issue." + "reason": "General v5 breakage and Qwen3_5 import failure are not the same underlying defect." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "RoPE implementation and audio positional encoding are distinct model components." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload mismatch and Gemma audio positional encoding are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:43704", + "accept": false, + "reason": "FP8 support request and VRAM leak are different topics." }, { "left": "issue:43704", "right": "issue:45341", "accept": false, - "reason": "VRAM leak versus a testing_utils bug; unrelated." + "reason": "VRAM leak and a testing_utils bug do not appear to share code." }, { "left": "issue:43352", "right": "issue:46218", "accept": false, - "reason": "Flash-attention support error versus missing model import; different code paths." + "reason": "Flash Attention support missing and module import failure are different problems." }, { - "left": "issue:44351", - "right": "issue:46218", + "left": "issue:44485", + "right": "issue:47177", "accept": false, - "reason": "Two import errors, but for different symbols and likely different causes." + "reason": "RoPE implementation bug and archive traversal vulnerability are unrelated." }, { - "left": "issue:44861", + "left": "issue:44829", "right": "issue:45290", "accept": false, - "reason": "Tied-weights crash versus chat-template tool-call crash; unrelated." + "reason": "Flash_attention_3 training degeneration and chat template tool-call crash are unrelated." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "FP8 support request versus processor save_pretrained file emission bug; unrelated." + "reason": "FP8 inference support and processor save_pretrained missing files are not the same issue." }, { "left": "issue:28348", "right": "issue:43352", "accept": false, - "reason": "Both concern Flash Attention 2.0, but they are different model-specific support requests/failures." + "reason": "Both are Flash Attention 2.0 support requests, but for different model classes and not clearly one fix." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs bug versus an irrelevant/non-technical issue; no relation." + "reason": "Scheduler kwargs bug and unrelated placeholder issue." }, { "left": "issue:44998", "right": "issue:47177", "accept": false, - "reason": "No meaningful technical overlap; unrelated issues." + "reason": "Noise issue versus security vulnerability; no overlap." }, { "left": "issue:44998", "right": "issue:47176", "accept": false, - "reason": "No meaningful technical overlap; unrelated issues." + "reason": "Noise issue versus shell-operator bypass vulnerability." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "FP8 support request versus a Qwen3.5 MoE from_pretrained error; different problems." + "reason": "FP8 support request and Qwen3.5 MoE from_pretrained error are different problems." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "Cache strategy typo versus unrelated issue; no duplicate signal." + "reason": "Typo in cache strategies and unrelated placeholder issue." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions regression versus unrelated issue." + "reason": "CLIPOutput attentions regression and unrelated placeholder issue." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Slow from_pretrained loading versus an import error; different failure mode and fix." + "reason": "Slow from_pretrained loading and import failure of layer_type_validation are unrelated." }, { "left": "issue:29870", "right": "issue:31515", "accept": false, - "reason": "BART generate decoder mask crash versus slow checkpoint loading; unrelated." + "reason": "BART generate UnboundLocalError and slow checkpoint loading are different bugs." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Continuous batching audio support versus fp8 inference support; different feature gaps." - }, - { - "left": "issue:30333", - "right": "issue:31515", - "accept": false, - "reason": "MLflow reporting failure versus checkpoint loading slowness; unrelated." - }, - { - "left": "issue:31515", - "right": "issue:36296", - "accept": false, - "reason": "Loading performance issue versus tensor-parallel training bug; unrelated." - }, - { - "left": "issue:31515", - "right": "issue:31795", - "accept": false, - "reason": "Runtime loading performance versus documentation confusion; unrelated." + "reason": "Audio continuous batching support and FP8 VL/MoE support are different feature gaps." } ] }, @@ -20650,6 +20573,10 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-540", "nodes": [ + "issue:30333", + "issue:31515", + "issue:31795", + "issue:36296", "issue:37934", "issue:42936", "issue:43240", @@ -20661,6 +20588,9 @@ "issue:46144" ], "soft_pairs": [ + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:31515|issue:31795", "issue:43717|issue:46144", "issue:42936|issue:46018", "issue:43668|issue:44031", @@ -20674,106 +20604,124 @@ "issue:37934|issue:43668", "issue:42936|issue:43668" ], - "prepared_review_unit_hash": "440478fe7ffb2b4fbe7e5b5176f66a390bb5e655ad6d4fbb81a33af8ab68f387", + "prepared_review_unit_hash": "d20f50b41f79c41bd21f5bcf13cd297b3115d204b9dc367eede63343c88eaa18", "component_budget": { "node_count": 540, "item_count": 540, - "soft_pair_count": 1994, - "serialized_chars": 1033373, - "estimated_input_tokens": 258344, - "estimated_eval_tokens": 516944 + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 }, "budget": { - "node_count": 9, - "item_count": 9, - "soft_pair_count": 12, - "serialized_chars": 8608, - "estimated_input_tokens": 2152, - "estimated_eval_tokens": 4560 + "node_count": 13, + "item_count": 13, + "soft_pair_count": 15, + "serialized_chars": 11766, + "estimated_input_tokens": 2942, + "estimated_eval_tokens": 6140 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a set of unrelated bug reports across different models/configs/tokenizers. The similarity signals are superficial; none of the soft pairs look like the same underlying defect.", - "confidence": 0.96, - "canonical_issue_reason": "If a single representative is needed, issue 46144 is the most concrete code-path bug and the broadest implementation concern, but it is not a true duplicate anchor for the rest of the cluster.", + "summary": "This cluster is a loose collection of unrelated issue reports across training, loading, docs, tokenization, model architecture, and configuration. None of the soft pairs look like the same underlying bug or change, so I would not merge any of them as duplicates.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 43240 is the most concrete and broadly actionable report in the set, with the most inbound references; if one issue must anchor this cluster, it is the best fit.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46144 is the strongest standalone report: it names a specific model, a specific attention implementation problem, and a plausible fix path. It is the best representative issue in this otherwise heterogeneous set.", + "best_issue_reason": "Issue 43240 is the strongest representative: it describes a clear code-path bug rather than a typo/docs-only concern, and it appears to have the most external attention.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43717", - "right": "issue:46144", + "left": "issue:30333", + "right": "issue:31515", "accept": false, - "reason": "Different models and different bugs: Mamba/Mamba-2 init_weights handling versus RoFormer attention interface usage." + "reason": "MLflow job-status reporting bug vs. slow from_pretrained checkpoint loading; different subsystems and symptoms." }, { - "left": "issue:42936", - "right": "issue:46018", + "left": "issue:31515", + "right": "issue:36296", "accept": false, - "reason": "Mask2Former ignore_value plumbing is unrelated to DeepSeek-V4 shared expert gating." + "reason": "Checkpoint load performance issue vs tensor-parallel training bug; not the same failure mode." }, { - "left": "issue:43668", - "right": "issue:44031", + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Loading speed complaint vs confusing forward docs; one is runtime performance, the other documentation." + }, + { + "left": "issue:43717", + "right": "issue:46144", + "accept": false, + "reason": "Mamba/Mamba-2 init_weights behavior vs RoFormer attention interface usage; unrelated model implementations and code paths." + }, + { + "left": "issue:42936", + "right": "issue:46018", + "accept": false, + "reason": "Mask2former ignore_value handling vs DeepSeek-V4 shared expert gating; different models and bugs." + }, + { + "left": "issue:43668", + "right": "issue:44031", "accept": false, - "reason": "One is a ModernBERT config type-hint issue; the other is a tokenizer regex warning regression." + "reason": "Incorrect type hint in ModernBERTConfig vs tokenizer regex warning after version change; unrelated issues." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Mask2Former ignore_value usage and loss kwargs forwarding are different code paths and different defects." + "reason": "Both concern model/loss plumbing, but one is a specific ignore_value usage bug and the other is kwargs not being forwarded to loss calculation; not the same underlying bug." }, { "left": "issue:43240", "right": "issue:46144", "accept": false, - "reason": "Loss-function kwargs propagation is unrelated to RoFormer attention interface behavior." + "reason": "Loss-kwargs forwarding bug vs RoFormer attention-interface issue; different code paths." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "Model ignore_value handling and tokenizer regex warnings are unrelated subsystems." + "reason": "Mask2former loss parameter bug vs tokenizer regex warning; no overlap in component or failure." }, { "left": "issue:42936", "right": "issue:46144", "accept": false, - "reason": "Different model families and no shared failing code path." + "reason": "Different models and different behaviors: ignore_value handling vs attention interface usage." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "Mask2Former loss/input handling is unrelated to a typo in quantization conversion naming." + "reason": "Mask2former loss behavior vs a typo in quantization_operation text; unrelated." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "ModernBERT config typing and quantization-typo reports do not describe the same bug." + "reason": "ModernBERT type-hint mistake vs quantization typo; both are minor metadata/text issues but not the same one." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "Loss kwargs forwarding versus a config annotation mistake are not the same underlying issue." + "reason": "Loss kwargs forwarding bug vs configuration type-hint error; separate bug classes." }, { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Llama4 norm implementation semantics and ModernBERT config type hints are unrelated." + "reason": "Question about Llama4TextL2Norm semantics vs ModernBERT config typing; unrelated model details." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2Former ignore_value usage is unrelated to ModernBERT norm_eps typing." + "reason": "Model loss parameter handling vs config type annotation; not the same issue." } ] }, @@ -20806,52 +20754,52 @@ "issue:10519|issue:11005", "issue:11005|issue:11307" ], - "prepared_review_unit_hash": "716c504e073322508ff6b24aac2a82db3abeac4cbfe78467c814e6f10b9ea577", + "prepared_review_unit_hash": "47bb4026a7f0071b2f0c56946b80bdcff6b639e434cf50ec83bc65c0b3b5df0c", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2877, - "estimated_input_tokens": 720, - "estimated_eval_tokens": 1696 + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2877, - "estimated_input_tokens": 720, - "estimated_eval_tokens": 1696 + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three issues are unrelated: one requests truncation from the beginning for tokenization, one asks for ReduceLROnPlateau-like scheduler behavior, and one asks for word time offsets in Wav2Vec2. No duplicate cluster is present.", + "summary": "These three issues are not duplicates: one is about tokenizer truncation behavior, one about ReduceLROnPlateau-style scheduling, and one about Wav2Vec2 word offset extraction. They do not describe the same underlying bug or feature request.", "confidence": 0.98, - "canonical_issue_reason": null, + "canonical_issue_reason": "No single canonical issue fits this cluster; the issues cover unrelated areas and solve different problems.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "No issue is a good global representative because the cluster is heterogeneous rather than a duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:10519", "right": "issue:11307", "accept": false, - "reason": "Different feature areas and problems: token truncation behavior vs Wav2Vec2 word timestamp extraction." + "reason": "Tokenizer truncation options and Wav2Vec2 word time offsets are unrelated feature requests with different code paths." }, { "left": "issue:10519", "right": "issue:11005", "accept": false, - "reason": "Unrelated requests: tokenizer truncation direction vs learning-rate scheduler functionality." + "reason": "Truncation-from-beginning and ReduceLROnPlateau-like scheduling address completely different components." }, { "left": "issue:11005", "right": "issue:11307", "accept": false, - "reason": "Different subsystems and use cases: scheduler behavior vs Wav2Vec2 alignment/timing offsets." + "reason": "Learning-rate scheduling functionality and speech word offset timing are distinct, non-overlapping issues." } ] }, @@ -20883,7 +20831,7 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "d0d20fb7464430aed0a10f27b7422b9ab5d4aa59c76e71db79a870ecec4b16bf", + "prepared_review_unit_hash": "8552eca202f87740a5af003b4072b6d2bfd5d3f2b4818dd2c13143fd59ac31af", "component_budget": { "node_count": 3, "item_count": 3, @@ -20905,24 +20853,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three issues do not look like duplicates of one another. They cover three different problems: Tapas tokenization mismatch, private-model download/auth error handling, and a generic bert-base-uncased loading failure.", - "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue fits the cluster well, because the reports describe different underlying bugs rather than one shared failure mode.", + "summary": "These three issues are all in the Transformers ecosystem, but they describe different problems: TAPAS tokenization mismatch, private-model auth/error messaging, and a generic BERT model-loading failure. I would not merge any pair as duplicates; issue 16618 is the broadest representative of the cluster\u2019s loose model-loading theme.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 16618 is the most general of the three and best matches the cluster\u2019s broad model-loading theme; the other two are narrower and clearly about different failure modes.", "canonical_pr_reason": null, - "best_issue_reason": "If one issue had to be kept as the most representative, issue #16618 is the broadest and most user-facing model-loading report, but it should not absorb the others.", + "best_issue_reason": "16618 is the strongest representative because it is a user-facing model load failure, whereas 13244 is tokenizer-specific and 15980 is specifically about private-model download/auth messaging.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Different bug classes: Tapas tokenization behavior vs failure to load bert-base-uncased. Same general ML area, but not the same code-path problem." + "reason": "Reject: TAPAS tokenization differences are a tokenizer/implementation issue, while 16618 is a generic failure to load a BERT model. They are not the same underlying bug." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Private-model download/auth error messaging is a different underlying issue from a public model loading failure; they would need different fixes." + "reason": "Reject: 15980 is about an incorrect error message when downloading a private model without login; 16618 is a separate model-loading failure for a public model. Different cause and fix." } ] }, @@ -20968,100 +20916,100 @@ "issue:22352|issue:25725", "issue:22372|issue:25725" ], - "prepared_review_unit_hash": "368597d888f1e556cdeb666f3ae60f7dd0c48f5f2a89a8074d332a759a3610d1", + "prepared_review_unit_hash": "28fd20179938c96502f9b30ebfa6ddbc7a036675d008116ab7939a4916426988", "component_budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 11, - "serialized_chars": 8191, - "estimated_input_tokens": 2048, - "estimated_eval_tokens": 4352 + "serialized_chars": 8195, + "estimated_input_tokens": 2049, + "estimated_eval_tokens": 4354 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 11, - "serialized_chars": 8191, - "estimated_input_tokens": 2048, - "estimated_eval_tokens": 4352 + "serialized_chars": 8195, + "estimated_input_tokens": 2049, + "estimated_eval_tokens": 4354 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are separate feature requests for adding different models or model-specific workflows. The shared wording is superficial (\u201cAdd X model\u201d), but the underlying changes target different architectures and use cases, so this is not a duplicate cluster.", - "confidence": 0.94, - "canonical_issue_reason": null, + "summary": "This cluster is a false positive: the issues are about different model additions or feature requests, not the same underlying bug/change. None of the soft edges should be merged.", + "confidence": 0.97, + "canonical_issue_reason": "No true canonical issue exists here; the items request different models/features (EDSR/MDSR, VATT, FlexiBERT, Restormer, Meta-Transformer, D_Nikud, etc.) and do not describe the same problem.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "No issue is a strong global representative for a duplicate set. If forced, issue 25725 ('Meta-Transformer') is the broadest title, but it is still a distinct request rather than a duplicate of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:19631", "right": "issue:22372", "accept": false, - "reason": "EDSR/MDSR and Restormer are different image-restoration models; they would require separate integrations and code paths." + "reason": "EDSR/MDSR and Restormer are different image restoration models; same broad domain, but not the same request." }, { "left": "issue:19631", "right": "issue:20362", "accept": false, - "reason": "EDSR/MDSR and FlexiBERT are unrelated model additions in different model families, so they are not the same change." + "reason": "EDSR/MDSR and FlexiBERT are unrelated model requests in different areas." }, { "left": "issue:19865", "right": "issue:25725", "accept": false, - "reason": "VATT and Meta-Transformer are distinct model architectures with different implementation requirements." + "reason": "VATT and Meta-Transformer are different architectures with no evidence of the same code path or change." }, { "left": "issue:19865", "right": "issue:22372", "accept": false, - "reason": "VATT and Restormer are different models for different tasks; no evidence they share the same concrete fix." + "reason": "VATT and Restormer are separate model additions, not duplicates." }, { "left": "issue:20334", "right": "issue:22352", "accept": false, - "reason": "longformer_content and Whisper XVector finetuning describe different workflows and targets; not the same underlying request." + "reason": "'longformer_content' and Whisper XVector finetuning are unrelated tasks/models." }, { "left": "issue:20334", "right": "issue:20362", "accept": false, - "reason": "longformer_content and FlexiBERT are separate model-related requests with no shared code-path problem." + "reason": "Longformer-content and FlexiBERT are different model requests." }, { "left": "issue:20334", "right": "issue:25725", "accept": false, - "reason": "longformer_content and Meta-Transformer are unrelated model additions; similarity is only topical." + "reason": "Longformer-content and Meta-Transformer target different models/features." }, { "left": "issue:20362", "right": "issue:25839", "accept": false, - "reason": "FlexiBERT and D_Nikud are distinct model requests, likely for different NLP tasks and implementations." + "reason": "FlexiBERT and D_Nikud are distinct model additions with no shared underlying bug/change." }, { "left": "issue:20362", "right": "issue:25725", "accept": false, - "reason": "FlexiBERT and Meta-Transformer are different model families and do not look mergeable into one PR." + "reason": "FlexiBERT and Meta-Transformer are unrelated model requests." }, { "left": "issue:22352", "right": "issue:25725", "accept": false, - "reason": "Whisper XVector finetuning and Meta-Transformer are not the same change; one is a training workflow, the other a model addition." + "reason": "Whisper XVector finetuning is not the same as adding Meta-Transformer." }, { "left": "issue:22372", "right": "issue:25725", "accept": false, - "reason": "Restormer and Meta-Transformer are unrelated model additions; no shared concrete bug or code-path." + "reason": "Restormer and Meta-Transformer are different models; no duplicate evidence beyond broad 'add model' similarity." } ] }, @@ -21159,12 +21107,12 @@ "issue:47299|issue:47402", "issue:46226|issue:47299" ], - "prepared_review_unit_hash": "5101f9e89a29c55e0620ba0ac3b15964448e67e7912f1285cc779a46ed77c18c", + "prepared_review_unit_hash": "03cc2ddfa384ca34997e8c0de08d525e48e3607602eb4f715a47e3bc935652e3", "component_budget": { "node_count": 16, "item_count": 16, "soft_pair_count": 55, - "serialized_chars": 29305, + "serialized_chars": 29306, "estimated_input_tokens": 7327, "estimated_eval_tokens": 14910 }, @@ -21172,7 +21120,7 @@ "node_count": 16, "item_count": 16, "soft_pair_count": 55, - "serialized_chars": 29305, + "serialized_chars": 29306, "estimated_input_tokens": 7327, "estimated_eval_tokens": 14910 }, @@ -21181,342 +21129,342 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly noisy similarity: one unrelated Flash Attention 2 vs GGUF pair, plus a large i18n translation-template family that shares boilerplate but appears to cover distinct language requests. I would not merge any of the soft pairs.", - "confidence": 0.95, - "canonical_issue_reason": "issue:43834 is the most generic hub among the i18n template issues, so it is the best anchor for that subgroup, though the overall cluster is not a single duplicate set.", + "summary": "This cluster is heterogeneous and appears to contain multiple unrelated themes: two architecture-support feature requests and a large set of i18n translation template issues. The soft similarities are mostly template-driven, not evidence of true duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "If forced to choose one issue as the anchor, issue 43834 is the most central representative of the dense i18n translation subgroup, but the overall cluster is not a true duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "Same as above: 43834 is the strongest representative of the near-identical translation-template subgroup, but not a global duplicate target for the unrelated items.", + "best_issue_reason": "Issue 43834 is the best single issue to represent the largest near-identical translation-template subgroup, though it should not be merged with the other language requests as duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Flash Attention 2 support and GGUF support are unrelated features." + "reason": "Different feature requests: Flash Attention 2 support vs GGUF support for more architectures." }, { "left": "issue:26794", "right": "issue:44370", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Both are i18n translation requests, but they appear to be separate language tasks rather than the same underlying issue." }, { "left": "issue:26794", "right": "issue:44365", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Same translation-program template, but no evidence it targets the same language or change." }, { "left": "issue:26794", "right": "issue:47402", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Translation-template similarity only; likely a different language request." }, { "left": "issue:26794", "right": "issue:43834", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Similar i18n template, but not enough evidence that these are the same translation task." }, { "left": "issue:26794", "right": "issue:47299", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different translation request; similar formatting is not enough to call them duplicates." }, { "left": "issue:26794", "right": "issue:27088", "accept": false, - "reason": "Different language requests; similarity comes from the shared i18n template." + "reason": "Kannada vs Turkish translation requests are distinct issues." }, { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Different language translation requests; only the template matches." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Template overlap only; no concrete shared translation target." }, { "left": "issue:27088", "right": "issue:47402", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Different i18n request, not the same underlying change." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Same translation template, but separate language-specific issues." }, { "left": "issue:28936", "right": "issue:43834", "accept": false, - "reason": "Different language request; shared wording is just the translation template." + "reason": "Spanish translation request vs generic i18n template; not clearly the same issue." }, { "left": "issue:27088", "right": "issue:28936", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Turkish and Spanish translation issues are different language tasks." }, { "left": "issue:28936", "right": "issue:44365", "accept": false, - "reason": "Different language request; shared wording is just the translation template." + "reason": "Different language request; similarity is only in the issue template." }, { "left": "issue:28936", "right": "issue:30240", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Spanish vs Polish translation requests are distinct." }, { "left": "issue:28936", "right": "issue:38435", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Different language translation issues; no evidence of duplication." }, { "left": "issue:27088", "right": "issue:30191", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Turkish translation vs Persian translation; separate issues." }, { "left": "issue:26794", "right": "issue:30191", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Kannada and Persian translation tasks are not the same change." }, { "left": "issue:30191", "right": "issue:30240", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Persian vs Polish translation requests are distinct language tasks." }, { "left": "issue:28936", "right": "issue:30191", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Spanish and Persian translation requests are separate issues." }, { "left": "issue:26794", "right": "issue:30240", "accept": false, - "reason": "Different language requests; same template, not the same task." + "reason": "Different target languages, so not duplicates." }, { "left": "issue:30240", "right": "issue:44370", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Polish translation request vs generic placeholder translation issue; no proof they are the same." }, { "left": "issue:30240", "right": "issue:44365", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Polish vs another i18n template issue; similar structure only." }, { "left": "issue:30240", "right": "issue:47402", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Different translation task, despite near-identical templated wording." }, { "left": "issue:30240", "right": "issue:43834", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Polish translation issue and a generic i18n placeholder are not enough to merge." }, { "left": "issue:38435", "right": "issue:44370", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Romanian translation request vs another i18n template issue; likely different language." }, { "left": "issue:38435", "right": "issue:44365", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Template overlap only; distinct translation requests." }, { "left": "issue:38435", "right": "issue:47402", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Romanian translation request is separate from the generic placeholder issue." }, { "left": "issue:38435", "right": "issue:43834", "accept": false, - "reason": "Different language request; the overlap is boilerplate translation text only." + "reason": "Different language tasks; not the same underlying issue." }, { "left": "issue:38435", "right": "issue:47299", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Romanian translation issue and the Thai-script issue are distinct." }, { "left": "issue:43834", "right": "issue:44370", "accept": false, - "reason": "Placeholder/template titles overlap, but there is no evidence they are the same language task." + "reason": "Both are templated i18n issues, but there is no evidence they refer to the same language." }, { "left": "issue:43834", "right": "issue:44365", "accept": false, - "reason": "Placeholder/template titles overlap, but there is no evidence they are the same language task." + "reason": "Near-identical template, but not enough to collapse separate translation requests." }, { "left": "issue:43834", "right": "issue:47402", "accept": false, - "reason": "Placeholder/template titles overlap, but there is no evidence they are the same language task." + "reason": "Same template family only; likely different language target." }, { "left": "issue:43834", "right": "issue:47299", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different underlying translation request, despite templated similarity." }, { "left": "issue:43834", "right": "issue:46226", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "The issues look templated, but no evidence they are the same language or task." }, { "left": "issue:44365", "right": "issue:44370", "accept": false, - "reason": "Placeholder/template titles overlap, but there is no evidence they are the same language task." + "reason": "Placeholder i18n titles only; not enough to show the same translation change." }, { "left": "issue:44365", "right": "issue:47402", "accept": false, - "reason": "Placeholder/template titles overlap, but there is no evidence they are the same language task." + "reason": "Same template, but likely different language requests." }, { "left": "issue:44365", "right": "issue:47299", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Distinct translation issue; template similarity is insufficient." }, { "left": "issue:44365", "right": "issue:46226", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "No concrete sign these are the same translation task." }, { "left": "issue:44370", "right": "issue:47402", "accept": false, - "reason": "Placeholder/template titles overlap, but there is no evidence they are the same language task." + "reason": "Templated i18n issues only; insufficient evidence of duplication." }, { "left": "issue:44370", "right": "issue:47299", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different translation request, not the same underlying issue." }, { "left": "issue:44370", "right": "issue:46226", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Same template family, but likely separate language tasks." }, { "left": "issue:46222", "right": "issue:47299", "accept": false, - "reason": "Different Thai-text request/title; not the same issue." + "reason": "Different-language/locale-style issues; similarity is only superficial." }, { "left": "issue:44370", "right": "issue:46222", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Templated i18n issue vs Thai-script issue; no evidence of the same request." }, { "left": "issue:44365", "right": "issue:46222", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different content despite superficial template similarity." }, { "left": "issue:46222", "right": "issue:47402", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Separate translation requests; not the same underlying change." }, { "left": "issue:43834", "right": "issue:46222", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "No indication these are the same language translation task." }, { "left": "issue:46226", "right": "issue:47402", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different issues with only templated resemblance." }, { "left": "issue:46226", "right": "issue:46931", "accept": false, - "reason": "Different Thai-text request/title; not the same issue." + "reason": "Distinct Thai-language issues; not enough to merge as duplicates." }, { "left": "issue:44370", "right": "issue:46931", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different translation/localization request; similarity is not concrete duplication." }, { "left": "issue:44365", "right": "issue:46931", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Template similarity only; likely separate topics." }, { "left": "issue:43834", "right": "issue:46931", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "The titles are generically similar, but the issues appear to be different requests." }, { "left": "issue:46931", "right": "issue:47402", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different translation issues; no evidence they are the same change." }, { "left": "issue:47299", "right": "issue:47402", "accept": false, - "reason": "Different language/request text; not the same underlying issue." + "reason": "Different language/localization issues, not the same underlying request." }, { "left": "issue:46226", "right": "issue:47299", "accept": false, - "reason": "Different Thai-text request/title; not the same issue." + "reason": "Separate issues with only superficial textual similarity." } ] }, @@ -21546,7 +21494,7 @@ "soft_pairs": [ "issue:30396|issue:31103" ], - "prepared_review_unit_hash": "c8272be11fae4419c3b25345337ed1250e4414e86bb1fde131850ed74403cf72", + "prepared_review_unit_hash": "5ff1df7fc293c7b99c7336f46903d747fb1c290894c97d33aa3f6bb72c125b26", "component_budget": { "node_count": 2, "item_count": 2, @@ -21568,24 +21516,27 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are related in theme but do not clearly describe the same underlying request: one asks to add a specific model, while the other asks to add a CausalLM class/implementation. I would not merge them as duplicates.", - "confidence": 0.71, - "canonical_issue_reason": "Issue 30396 is the more concrete feature request, with a specific model name and clearer scope.", + "summary": "The two issues are closely related HelpingAI requests, but the packet only provides titles and soft similarity. That is not enough to conclude they are the same underlying request; #31103 could be an implementation/integration task rather than the specific model-addition request in #30396.", + "confidence": 0.55, + "canonical_issue_reason": "#30396 is the clearer and more specific request title, so it is the better representative of the cluster if these are grouped at all.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 30396 is the best representative because it is more specific and self-contained.", + "best_issue_reason": "#30396 is the best issue to surface because it names the concrete HelpingAI-3B-v2.2 request more explicitly than #31103, while still avoiding overclaiming that the two are definite duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:30396", "right": "issue:31103", "accept": false, - "reason": "Both are about HelpingAI-related additions, but the requested change is not the same: one is a model add request, the other is an API/class addition request. Too little evidence they are the same underlying issue." + "reason": "The titles are similar, but we only have soft similarity and no body/link evidence. #31103 may be an implementation or integration task related to #30396 rather than the same exact request, so this edge should stay rejected." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary goes a bit beyond the packet evidence. We only have titles and a soft-similarity signal, so saying both issues are the same underlying feature request is stronger than warranted. In particular, issue 31103 (\u201cAdd HelpingAIForCausalLM to it\u201d) could be an implementation/integration task related to, but not necessarily the same as, the specific model request in 30396. The soft-edge verdict should stay more conservative unless the bodies or links confirm the match." + }, + "evaluator_used": true, + "retried": true, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -21688,370 +21639,388 @@ "issue:42503|issue:46653", "issue:45027|issue:46653" ], - "prepared_review_unit_hash": "b00711c22ecd7e1ca5063c3226844f96bebf6d26c8724f0e2a213900b8936288", + "prepared_review_unit_hash": "5cafd11c891e9c8332d2c75ac8473c79f8421e6ef7b923b1e76a3e1c20c98db6", "component_budget": { "node_count": 23, "item_count": 23, "soft_pair_count": 59, - "serialized_chars": 33613, - "estimated_input_tokens": 8404, - "estimated_eval_tokens": 17064 + "serialized_chars": 33618, + "estimated_input_tokens": 8405, + "estimated_eval_tokens": 17066 }, "budget": { "node_count": 23, "item_count": 23, "soft_pair_count": 59, - "serialized_chars": 33613, - "estimated_input_tokens": 8404, - "estimated_eval_tokens": 17064 + "serialized_chars": 33618, + "estimated_input_tokens": 8405, + "estimated_eval_tokens": 17066 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are mostly unrelated model-request tickets that share generic wording like \"add/support model\". The similarity is topical, not duplicative, so I would not collapse the cluster into one canonical change.", + "summary": "This cluster is heterogeneous: it mixes unrelated model requests across vision, audio, NLP, keypoint matching, and config/model-architecture updates. The soft similarities are mostly driven by generic 'add/support model' wording, not the same underlying change, so none should be deduplicated.", "confidence": 0.97, - "canonical_issue_reason": "No clear canonical issue: the cluster mixes many distinct model-addition requests across different architectures, checkpoints, and modalities.", + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a safe global representative; choosing one would incorrectly merge unrelated requests.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36718", "right": "issue:47004", "accept": false, - "reason": "Related keypoint tasks, but RoMa matcher and LoMa detector+matcher are different model requests." + "reason": "Both are keypoint-related, but RoMa matcher vs LoMa detector+matcher are different model requests and code paths." }, { "left": "issue:36718", "right": "issue:41115", "accept": false, - "reason": "Different scope: keypoint matcher request vs MiniCPM3 model architecture support." + "reason": "RoMa keypoint matching and MiniCPM3 architecture are unrelated model additions." }, { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different models entirely: EuroBert config vs DEIMv2 support." + "reason": "EuroBert config and DEIMv2 are different models with different integration work." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBert config and Qwen3-TTS are unrelated model requests." + "reason": "EuroBert config and Qwen3-TTS are different model families and modalities." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "Different model families: EuroBert vs RF-DETR." + "reason": "EuroBert config and RF-DETR model requests are unrelated." }, { "left": "issue:36817", "right": "issue:42954", "accept": false, - "reason": "Different requests: EuroBert config vs XiaoMi MiMo." + "reason": "EuroBert and XiaoMi MiMo are distinct model additions." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are different model additions." + "reason": "RF-DETR and DEIMv2 are different vision models, not the same change." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "Detection model vs audio model; not the same change." + "reason": "RF-DETR and LFM2.5 Audio target different modalities and code paths." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "RF-DETR support is unrelated to the weight-sparse transformer request." + "reason": "RF-DETR and OpenAI weight-sparse transformer are unrelated model requests." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR and Qwen3-ASR target different modalities and code paths." + "reason": "RF-DETR and Qwen3-ASR are different model families and tasks." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR support is unrelated to Qwen3-TTS support." + "reason": "RF-DETR vision detector request is unrelated to Qwen3-TTS audio support." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are different model requests." + "reason": "NeoBERT and Qwen3-ASR are unrelated model additions." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT support and the weight-sparse transformer request are unrelated." + "reason": "NeoBERT and circuit-sparsity/weight-sparse transformer are different architectures and requests." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT is a text model; LFM2.5 Audio is an audio model." + "reason": "NeoBERT and LFM2.5 Audio are different modalities and model families." }, { "left": "issue:37015", "right": "issue:37334", "accept": false, - "reason": "NeoBERT and LLMDet are different model additions." + "reason": "NeoBERT and LLMDet are distinct models; not the same underlying change." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR detection support and NeoBERT support are unrelated." + "reason": "RF-DETR and NeoBERT are unrelated model additions." }, { "left": "issue:37334", "right": "issue:41115", "accept": false, - "reason": "LLMDet support and MiniCPM3 architecture support are different changes." + "reason": "LLMDet support and MiniCPM3 architecture are different integration efforts." }, { "left": "issue:37334", "right": "issue:43837", "accept": false, - "reason": "LLMDet and Qwen3-ASR are different models with different code paths." + "reason": "LLMDet and Qwen3-ASR are different model requests." }, { "left": "issue:37334", "right": "issue:43671", "accept": false, - "reason": "LLMDet and Qwen3-TTS are unrelated model requests." + "reason": "LLMDet and Qwen3-TTS are unrelated model families and tasks." }, { "left": "issue:37334", "right": "issue:42503", "accept": false, - "reason": "LLMDet support is unrelated to ModernVBERT model support." + "reason": "LLMDet and ModernVBERT are different model additions." }, { "left": "issue:40330", "right": "issue:42503", "accept": false, - "reason": "open-qwen2vl-base and ModernVBERT are different model requests." + "reason": "open-qwen2vl-base and ModernVBERT are different models despite both being vision-language oriented." }, { "left": "issue:40330", "right": "issue:43671", "accept": false, - "reason": "Vision-language model support vs TTS support; not the same change." + "reason": "open-qwen2vl-base and Qwen3-TTS are different tasks and implementations." }, { "left": "issue:40330", "right": "issue:42954", "accept": false, - "reason": "open-qwen2vl-base and XiaoMi MiMo are unrelated models." + "reason": "open-qwen2vl-base and XiaoMi MiMo are unrelated requests." }, { "left": "issue:40330", "right": "issue:46653", "accept": false, - "reason": "open-qwen2vl-base is a model request; Mirko Privitera is unrelated to model support." + "reason": "open-qwen2vl-base and Mirko Privitera are clearly unrelated." }, { "left": "issue:40330", "right": "issue:42738", "accept": false, - "reason": "open-qwen2vl-base and BERT-like models with RoPE are different requests." + "reason": "open-qwen2vl-base and BERT-like models with RoPE are not the same change." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 support and Qwen3-ASR support are different model additions." + "reason": "MiniCPM3 architecture and Qwen3-ASR support are unrelated model requests." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 and the weight-sparse transformer request are unrelated." + "reason": "MiniCPM3 and the weight-sparse transformer request target different model families." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 support is unrelated to ModernVBERT models." + "reason": "MiniCPM3 architecture and ModernVBERT models are different additions." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are different models and modalities." + "reason": "DEIMv2 and Qwen3-TTS are unrelated models and modalities." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are both BERT-like, but they are distinct model requests." + "reason": "NeoBERT and ModernVBERT are distinct BERT-like model requests, not the same underlying change." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT and Qwen3-ASR are different model families and tasks." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like RoPE models and Voxtral TTS are unrelated model requests." + "reason": "BERT-like RoPE models and Voxtral TTS support are unrelated." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT and BERT-like models with RoPE are different model support requests." + "reason": "ModernVBERT and generic BERT-like RoPE support overlap in theme but not in concrete requested model or code path." }, { "left": "issue:42738", "right": "issue:42954", "accept": false, - "reason": "BERT-like RoPE models and XiaoMi MiMo are not the same change." + "reason": "RoPE BERT-like models and XiaoMi MiMo are different requests." }, { "left": "issue:42738", "right": "issue:46653", "accept": false, - "reason": "BERT-like RoPE support is unrelated to the Mirko Privitera issue." + "reason": "RoPE BERT-like models and Mirko Privitera are unrelated." }, { "left": "issue:42954", "right": "issue:44205", "accept": false, - "reason": "XiaoMi MiMo and SAM3-LiteText are different model requests." + "reason": "XiaoMi MiMo and SAM3-LiteText are different model additions." }, { - "left": "issue:42954", - "right": "issue:45177", + "left": "issue:37334", + "right": "issue:42954", "accept": false, - "reason": "XiaoMi MiMo and DeepSeek-OCR2 are unrelated model additions." + "reason": "LLMDet and XiaoMi MiMo are unrelated model requests." }, { "left": "issue:42954", - "right": "issue:46653", + "right": "issue:43671", "accept": false, - "reason": "XiaoMi MiMo support is unrelated to the Mirko Privitera issue." + "reason": "XiaoMi MiMo and Qwen3-TTS are different models and modalities." }, { "left": "issue:42503", "right": "issue:42954", "accept": false, - "reason": "ModernVBERT models and XiaoMi MiMo are different requests." + "reason": "ModernVBERT and XiaoMi MiMo are unrelated additions." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Qwen3-TTS and Qwen3-ASR are opposite speech tasks and should not be merged." + "reason": "Qwen3-TTS and Qwen3-ASR are sibling tasks but different concrete model support changes." }, { "left": "issue:43671", - "right": "issue:44121", + "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS support is unrelated to the weight-sparse transformer request." + "reason": "Qwen3-TTS and Voxtral TTS support are different TTS model requests, not the same change." }, { "left": "issue:43671", - "right": "issue:45027", + "right": "issue:45177", "accept": false, - "reason": "Both mention TTS, but Qwen3-TTS and Voxtral-4B-TTS are different model-specific requests." + "reason": "Qwen3-TTS and DeepSeek-OCR2 are different modalities and code paths." }, { "left": "issue:43671", - "right": "issue:45177", + "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS support and DeepSeek-OCR2 are unrelated." + "reason": "Qwen3-TTS and the weight-sparse transformer request are unrelated." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio are both audio-related, but they are different model requests." + "reason": "Qwen3-ASR and LFM2.5 Audio are different audio model requests." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "Qwen3-ASR support and EUPE are different model additions." + "reason": "Qwen3-ASR and EUPE are different model families." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "Qwen3-ASR support is unrelated to the weight-sparse transformer request." - }, - { - "left": "issue:43837", - "right": "issue:45027", - "accept": false, - "reason": "ASR vs TTS; Qwen3-ASR and Voxtral-4B-TTS are not the same fix." + "reason": "Qwen3-ASR and the weight-sparse transformer request are unrelated." }, { "left": "issue:43909", "right": "issue:44205", "accept": false, - "reason": "LFM2.5 Audio and SAM3-LiteText are different modality/model requests." + "reason": "LFM2.5 Audio and SAM3-LiteText are different modalities and integrations." }, { "left": "issue:43909", "right": "issue:46645", "accept": false, - "reason": "LFM2.5 Audio support and Granite SWA+sink are different code changes." + "reason": "LFM2.5 Audio and Granite SWA+sink option are unrelated changes." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "Weight-sparse transformer support and EUPE are unrelated model requests." + "reason": "OpenAI weight-sparse transformer and EUPE are different model requests." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 support and Voxtral-4B-TTS support are different model additions." + "reason": "MiniCPM3 and Voxtral TTS support are unrelated." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS support are different audio tasks and models." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS support are unrelated." + }, + { + "left": "issue:37334", + "right": "issue:45027", + "accept": false, + "reason": "LLMDet and Voxtral TTS support are unrelated." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "Qwen3-ASR and DeepSeek-OCR2 are different modalities and code paths." + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model requests." }, { "left": "issue:42954", - "right": "issue:43671", + "right": "issue:45177", "accept": false, - "reason": "XiaoMi MiMo and Qwen3-TTS are unrelated model requests." + "reason": "XiaoMi MiMo and DeepSeek-OCR2 are unrelated." }, { "left": "issue:45322", "right": "issue:46133", "accept": false, - "reason": "EUPE and TIPSv2 are both model requests, but they are distinct models and changes." + "reason": "EUPE and TIPSv2 are both research-model requests but they are different architectures and not the same change." }, { "left": "issue:42954", "right": "issue:46653", "accept": false, - "reason": "XiaoMi MiMo support is unrelated to the Mirko Privitera issue." + "reason": "XiaoMi MiMo and Mirko Privitera are unrelated." }, { "left": "issue:42503", "right": "issue:46653", "accept": false, - "reason": "ModernVBERT models are unrelated to the Mirko Privitera issue." + "reason": "ModernVBERT and Mirko Privitera are unrelated." }, { "left": "issue:45027", "right": "issue:46653", "accept": false, - "reason": "Voxtral-4B-TTS support is unrelated to the Mirko Privitera issue." + "reason": "Voxtral TTS support and Mirko Privitera are unrelated." } ] }, @@ -22081,12 +22050,12 @@ "soft_pairs": [ "issue:38740|issue:46443" ], - "prepared_review_unit_hash": "db8f1fdb92f238068d07ea61086e63748b6470b06e0ee593e9b741e478236c47", + "prepared_review_unit_hash": "ff253ac42ef146eecdef940b866acab7b560f5c710b0b89b8bc205c74141006b", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1506, + "serialized_chars": 1505, "estimated_input_tokens": 377, "estimated_eval_tokens": 1010 }, @@ -22094,7 +22063,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1506, + "serialized_chars": 1505, "estimated_input_tokens": 377, "estimated_eval_tokens": 1010 }, @@ -22103,18 +22072,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are unrelated: one is a docs request to add Pruna as an optimization framework, the other is a vision RoPE refactor. No duplicate relationship is supported.", + "summary": "These two issues are not duplicates: one is a documentation request to add Pruna as an optimization framework, while the other is a technical Vision RoPE refactor. They describe different domains and different kinds of work.", "confidence": 0.98, - "canonical_issue_reason": "Issue #46443 is the more substantive technical item and is the only one plausibly central to a code change; #38740 is a standalone docs request and unrelated.", + "canonical_issue_reason": "Neither issue is a clear canonical duplicate of the other; they are unrelated topics.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #46443 is the better representative issue for this set because it describes an actual implementation refactor, while #38740 is a separate documentation task.", + "best_issue_reason": "Issue #46443 is the better standalone representative because it is a concrete technical refactor, whereas #38740 is only a docs request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:38740", "right": "issue:46443", "accept": false, - "reason": "Different topics and scopes: docs/third-party framework mention versus a vision RoPE refactor. They do not appear to describe the same underlying bug or change." + "reason": "Different subject matter: documentation/integration request versus Vision RoPE refactor. No shared code path or duplicate bug/change." } ] }, @@ -22144,12 +22113,12 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "326d4733cdb4c6d1eefbc00a609bc0fe7ea52438fec0b6b00c7ee37f5211ff46", + "prepared_review_unit_hash": "805f5e6ae6c5faac6369a1ab9f78c2ba41f4c3adf5f109176fe01434bb549a60", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, + "serialized_chars": 1479, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -22157,7 +22126,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, + "serialized_chars": 1479, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -22166,8 +22135,8 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues describe different requests: one is about block decoding behavior, the other about adding DeepSeek v3.2 support. There is no evidence they are duplicates or the same underlying change.", - "confidence": 0.97, + "summary": "The two issues are not duplicates. \"Set Block Decoding\" is about a decoding behavior/code-path change, while \"DeepSeek v3.2 support\" is a model-specific support request. The soft similarity is superficial only.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, @@ -22177,7 +22146,7 @@ "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "Different scope and target: block decoding is a decoding-mode feature, while DeepSeek v3.2 support is a model-compatibility request." + "reason": "Different underlying requests with no clear shared bug or change: block decoding behavior vs support for a specific model version." } ] }, @@ -22207,40 +22176,40 @@ "soft_pairs": [ "issue:41306|issue:41826" ], - "prepared_review_unit_hash": "ec7d6b44e10ff3e9df203d462642c2b468cb2912cd50aa1a5430e3e769a5413b", + "prepared_review_unit_hash": "7d815a7c8c225ccd997c83281b525450975b799b60552b007900e98e34e26afd", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, - "estimated_input_tokens": 396, - "estimated_eval_tokens": 1048 + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, - "estimated_input_tokens": 396, - "estimated_eval_tokens": 1048 + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are both about efficiency-oriented model changes, but they target different mechanisms and code paths: FLCE flag handling in `from_pretrained()` versus integrating TiledMLP for lower memory use. They do not look like duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "No clear canonical issue for this pair; they describe different feature requests in different model components rather than the same underlying bug or change.", + "summary": "These are separate enhancement issues: one requests a `from_pretrained()` flag for TiledFusedLogitsLoss/FLCE, while the other asks for integrating TiledMLP to reduce memory footprint. They target different components and underlying changes.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 41306 is the more concrete, narrowly scoped request about exposing a specific `from_pretrained()` flag for FLCE/TiledFusedLogitsLoss.", "canonical_pr_reason": null, - "best_issue_reason": "If one issue must be picked as the best representative, #41306 is more concrete and implementation-bound, with a specific API/behavior target (`from_pretrained()` flag handling).", + "best_issue_reason": "Issue 41306 is the better representative because it has a clear API change request and is more self-contained than the broader TiledMLP integration idea.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41306", "right": "issue:41826", "accept": false, - "reason": "Different requests and likely different code paths: one is about TiledFusedLogitsLoss/FLCE configuration in `from_pretrained()`, the other is about adding TiledMLP to reduce memory footprint. Shared high-level efficiency theme is too broad to treat as the same issue." + "reason": "Different underlying changes: one is an API flag for loading a specific loss module, the other is a separate memory-optimization integration for TiledMLP. They are not the same bug or change." } ] }, @@ -22270,7 +22239,7 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "ff40b263f58b92ef6683d79ab88d0b68e543b1042d00784d024cfea6924009d9", + "prepared_review_unit_hash": "2f974ed6de37fd502ef86a0974cd99d68f65403edcd22ff7e38d07dd8d9aa2a0", "component_budget": { "node_count": 2, "item_count": 2, @@ -22292,18 +22261,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are related only at a very high level around Trainer customization, but they describe different problems: unused-column handling vs FLOPS estimation for custom trainers. They should not be treated as duplicates.", - "confidence": 0.91, - "canonical_issue_reason": "No single canonical issue: the reports target different code paths and user-visible behaviors.", + "summary": "Both issues are about Trainer extensibility, but they describe different bugs: one is about `remove_unused_column` behavior, the other about FLOPS estimation for custom trainers. They are related, not duplicates.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 43598 is the better representative because it targets a concrete, user-facing Trainer behavior change (`remove_unused_column`) with a clear code path to adjust.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43598 is the more central/reusable representative of the cluster because it focuses on Trainer behavior that affects customization broadly; issue 43600 is a separate estimation concern.", + "best_issue_reason": "Issue 43598 is slightly stronger as the canonical issue because it is more specific and actionable; 43600 is a separate accounting/metrics problem.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Different underlying problems: one is about `remove_unused_column` behavior/customizability, the other about FLOPS estimation for custom trainers. Shared Trainer context is too broad to merge." + "reason": "Different underlying problems: one concerns automatic column removal in Trainer, the other concerns FLOPS counting for custom trainers. They share the Trainer/customization theme but are not the same fix." } ] }, @@ -22333,7 +22302,7 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "ee752b1f58376d6eb0f4c559f0c10f1ea34f68f7e6167b25bb224b1566a9e67a", + "prepared_review_unit_hash": "185c61c718d6759947d1a5ae4055f36cffe5bbb86191d90cd467c15145281425", "component_budget": { "node_count": 2, "item_count": 2, @@ -22355,18 +22324,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are separate feature requests for different Qwen model families and different AutoModel task heads, so they should not be treated as duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "Issue 43684 is the broader model-registration request, but it does not subsume 44405; the two issues target different configs and heads.", + "summary": "These are related only at a very high level: both ask for additional AutoModel support for Qwen variants, but they target different model families and different integration points. They should not be treated as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 43684 is the broader request about registering Qwen3-Omni with AutoModel/AutoModelForConditionalGeneration, so it is the better representative of the general auto-model registration theme.", "canonical_pr_reason": null, - "best_issue_reason": "43684 is the closest representative of the cluster only because it is the more general AutoModel registration request.", + "best_issue_reason": "43684 is the most general and central issue in this pair; 44405 is a separate, more specific request for AutoModelForSequenceClassification on Qwen3.5.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different underlying requests: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration vs. Qwen3.5 sequence-classification support for AutoModelForSequenceClassification. Same broad area, but not the same fix." + "reason": "Different underlying change requests: Qwen3-Omni auto-registration for generation vs Qwen3.5 sequence-classification support. Same broad subsystem, but not the same bug or mergeable fix." } ] }, @@ -22403,12 +22372,12 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "a371a5aca064f6c98c7bd60536850f97f4c6e255d084ef548f41336877500e92", + "prepared_review_unit_hash": "579ff5b772d133b7198f701664453049b4c2fb926cd0bf1491b931b6e28f11bd", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4262, + "serialized_chars": 4264, "estimated_input_tokens": 1066, "estimated_eval_tokens": 2388 }, @@ -22416,7 +22385,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4262, + "serialized_chars": 4264, "estimated_input_tokens": 1066, "estimated_eval_tokens": 2388 }, @@ -22425,48 +22394,48 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "A tiny cluster of four very low-information issues: one off-topic-looking title (\u201cFett\u201d) and three nearly identical \u201chuggingface \u201d issues. There isn\u2019t enough substantive content to confidently merge any pair, though 43845/43846 are the closest match by title/body similarity.", - "confidence": 0.72, - "canonical_issue_reason": "Issue 43845 is the earliest of the repeated \u201chuggingface \u201d reports and has the strongest duplicate signal among the set, so it is the best placeholder canonical issue for this sparse cluster.", + "summary": "These are extremely sparse issue reports: one unrelated title ('Fett') and three near-identical 'huggingface ' issues with no actionable body detail. There is not enough evidence to confidently confirm true duplicates beyond superficial title similarity.", + "confidence": 0.34, + "canonical_issue_reason": "issue:43845 is the earliest of the repeated 'huggingface ' reports and the cleanest anchor for the cluster, but the report text is too thin to establish more than superficial similarity.", "canonical_pr_reason": null, - "best_issue_reason": "43845 is the best issue representative because it is one of the repeated huggingface-titled reports, appears earliest, and is more plausibly the anchor for deduping the nearly identical follow-up issue 43846.", + "best_issue_reason": "issue:43845 is the best representative issue because it is an early instance of the repeated title pattern; however, all of the issues are too under-described to be strong canonical candidates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles and no body/detail evidence tying them to the same problem." + "reason": "Different titles ('Fett' vs 'huggingface ') and no body detail or shared targets to show the same underlying bug." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles and no body/detail evidence tying them to the same problem." + "reason": "Different titles and no substantive content linking them to the same problem." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no body/detail evidence tying them to the same problem." + "reason": "Different titles and no actionable description to justify a duplicate relationship." }, { "left": "issue:43845", "right": "issue:43846", "accept": false, - "reason": "Very similar titles, but the issues are too content-poor to verify they describe the same underlying bug." + "reason": "The titles are the same, but both reports are too vague to confirm the same concrete issue with confidence." }, { "left": "issue:43845", "right": "issue:43859", "accept": false, - "reason": "Same broad title, but not enough substantive description to confirm a duplicate." + "reason": "Only superficial title similarity is present; the bodies are too sparse to prove the same bug." }, { "left": "issue:43846", "right": "issue:43859", "accept": false, - "reason": "Same broad title, but not enough substantive description to confirm a duplicate." + "reason": "Only repeated generic title text is shared; there is no evidence of the same underlying issue." } ] }, @@ -22511,106 +22480,106 @@ "issue:44573|issue:47036", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "39ec439a241b2e018af292f1abaefda54bc2305c01a014b853f5e238233fd404", + "prepared_review_unit_hash": "2cb9c05df8cf46a86091e92c74ac9b80e9f4d9448f1511c77a578075d0e26401", "component_budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 12, - "serialized_chars": 7361, - "estimated_input_tokens": 1841, - "estimated_eval_tokens": 3938 + "serialized_chars": 7355, + "estimated_input_tokens": 1839, + "estimated_eval_tokens": 3934 }, "budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 12, - "serialized_chars": 7361, - "estimated_input_tokens": 1841, - "estimated_eval_tokens": 3938 + "serialized_chars": 7355, + "estimated_input_tokens": 1839, + "estimated_eval_tokens": 3934 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly spam/off-topic issues with no clear shared bug or feature request. I reject all soft edges; the only exact-title matches are still too underspecified to treat as the same underlying actionable issue.", - "confidence": 0.95, + "summary": "These are mostly spam/generic placeholder issues with no concrete shared bug or change. The pairwise similarity is superficial and does not justify deduplication; no soft edges should be merged.", + "confidence": 0.93, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:44418 is the least-noisy and most actionable-looking item, but it is still too vague to serve as a real canonical issue.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are just '' placeholders with no actionable details; not enough evidence they represent the same underlying bug or change." + "reason": "Both are spam-like placeholders with no describable underlying problem; no concrete duplicate signal." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Same as above: identical spam-like titles, but no concrete shared problem to collapse into one issue." + "reason": "Same generic spam pattern; no evidence they report the same issue." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "'' and 'Documentation Enhancement Suggestion' are unrelated and describe different kinds of content." + "reason": "One is spam, the other is a vague documentation suggestion; not the same underlying bug/change." }, { "left": "issue:44371", "right": "issue:47036", "accept": false, - "reason": "No clear relationship between a spam placeholder and 'Community benchmark tip'." + "reason": "Spam placeholder vs. vague community tip; no shared concrete problem." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "No shared bug/change; 'Check out \"Google LLC\"' is unrelated to the spam placeholder." + "reason": "Spam placeholder vs. an unrelated titled issue; similarity is too weak for a duplicate." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "A documentation enhancement request is not the same underlying issue as a spam placeholder." + "reason": "Generic documentation request and spam placeholder are not the same change." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "Different content and intent; no evidence of the same actionable issue." + "reason": "No concrete overlap beyond being low-information issues." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Exact same placeholder title, but the bodies are too empty to confirm a real duplicate bug/change; treat as spam duplicates rather than a mergeable issue pair." + "reason": "Both are spam-like placeholders, but there is no evidence of a shared underlying bug or requested change." }, { "left": "issue:44572", "right": "issue:47036", "accept": false, - "reason": "No meaningful overlap between a spam placeholder and a community benchmark tip." + "reason": "Spam placeholder vs. community benchmark tip; too vague to merge." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Unrelated topics; nothing indicates the same underlying issue." + "reason": "No concrete shared issue or code path is evident." }, { "left": "issue:44573", "right": "issue:47036", "accept": false, - "reason": "No shared concrete problem or change request." + "reason": "Generic placeholder and benchmark tip are not duplicates." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "No evidence these are the same issue; titles point to different kinds of off-topic content." + "reason": "No substantive evidence of the same underlying problem." } ] }, @@ -22640,7 +22609,7 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "687660fa16ae3ce49a1c257375fd5351114fa2aa6c87cc536971ca2f6c5eaa0f", + "prepared_review_unit_hash": "f08750887a30026e629dd9363cb31605d6cebf6f3797312257b92a9cff0998c1", "component_budget": { "node_count": 2, "item_count": 2, @@ -22662,27 +22631,27 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues have identical titles and very similar body length/content signals, so they appear to be duplicates of the same report/request. Issue #45145 is the better canonical issue because it was created first and has an inbound reference; there are no PRs in this cluster.", - "confidence": 0.98, - "canonical_issue_reason": "#45145 is the earlier report and has the only inbound reference, so it is the best canonical issue for this duplicate pair.", + "summary": "These closed issues look like likely duplicate / very similar reports: they share the same title, have similar body lengths, and there is no filenames/targets metadata to distinguish them. #45145 is the best canonical issue because it was opened earlier and already has an inbound reference.", + "confidence": 0.86, + "canonical_issue_reason": "Issue #45145 is the better canonical issue because it was opened slightly earlier and already has an inbound reference, making it the most natural anchor for a likely-duplicate issue pair.", "canonical_pr_reason": null, - "best_issue_reason": "#45145 is the strongest representative issue: same title as #45151, earlier creation time, and more linkage evidence.", + "best_issue_reason": "Issue #45145 is the best issue representative because it is the earliest report and already has an inbound reference; the other issue appears to be a very similar duplicate report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", "accept": true, - "reason": "Same title and nearly identical issue metadata/body size strongly indicate the same underlying issue; this is a straightforward duplicate." + "reason": "Same title and very similar report size, with no differing filenames or explicit targets; this supports treating them as likely duplicate / very similar reports, though not as proven identical." } ] }, "evaluator_result": { - "accept": true, - "feedback": "Grounded in the packet: the issues have identical titles, close body lengths, both are closed, and #45145 is earlier with the only inbound reference. The duplicate/canonical conclusions are conservative enough, and there are no PRs to evaluate." + "accept": false, + "feedback": "The packet supports a likely duplicate relationship, but the writeup is too strong in places. We only have same title, similar body length, no filenames/targets, and a soft-similarity signal; we do not have direct evidence that the bodies are \u201ceffectively identical\u201d or that the content is \u201chighly overlapping.\u201d Please downgrade the summary and soft-edge language to something more conservative like \u201clikely duplicate / very similar reports.\u201d" }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -22760,7 +22729,7 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "a489de44f04c73b636836a5aff1c801008226b38fd9e7fdadb91392ea4bcd1a3", + "prepared_review_unit_hash": "10aaabfe4079087e968315c4156f28829abc0088e84080201515fd6e449b0849", "component_budget": { "node_count": 2, "item_count": 2, @@ -22782,18 +22751,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Reject as duplicate: both closed issues are empty, name/date-style titles with no body or references, and there is no evidence they report the same underlying problem.", - "confidence": 0.95, + "summary": "Two closed issues with unrelated person-name titles and no body or linkage; they do not look like duplicates of the same underlying bug or request.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "No issue is a good canonical candidate; both are content-free and appear unrelated/spam-like rather than duplicate bug reports.", - "best_pr_reason": null, + "best_issue_reason": "No issue is a meaningful canonical representative here because the two issues are unrelated and appear to be non-technical spam/placeholder entries.", + "best_pr_reason": "No PRs are present in this cluster.", "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "Shared surname similarity is not enough; the titles name different people with different dates and provide no bug/context overlap, so they do not look like the same issue." + "reason": "Different names and dates, no shared context, body, or explicit linkage; not the same underlying issue." } ] }, @@ -22823,40 +22792,40 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "4a83f3a76e75d7e33acefea8062e5bd1f5c908d8623a812dfb12a89a1412e5e8", + "prepared_review_unit_hash": "0b7efa2185c5fbfe0024b6a94a19276470f6603c78b6a6abbc8798ae65b5ed1e", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, - "estimated_input_tokens": 370, - "estimated_eval_tokens": 996 + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, - "estimated_input_tokens": 370, - "estimated_eval_tokens": 996 + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two closed issues are not duplicates; they are just two unrelated titles containing similar personal-name/date patterns. No PRs are present.", + "summary": "The two closed issues look like unrelated spam-like entries: each is just a different personal name plus a date, with no bug-report context. They should not be treated as duplicates.", "confidence": 0.98, - "canonical_issue_reason": null, + "canonical_issue_reason": "No canonical issue: the items have different names/dates and no shared problem description.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "No strong best representative; if forced, 45764 is only marginally earlier, but neither issue is a meaningful cluster anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "Different subject names and dates; no shared problem statement, code path, or actionable linkage. The similarity is only superficial." + "reason": "Reject: different personal-name/date strings with no evidence of the same underlying bug or change." } ] }, @@ -22886,7 +22855,7 @@ "soft_pairs": [ "issue:45810|issue:46295" ], - "prepared_review_unit_hash": "93a9e807825895d5c043c0ce53063d84aac2ddd7d36c7e2a696890cba8d5706e", + "prepared_review_unit_hash": "159856e189840e7da83d431707305e047e4210335189f2735c48a10d8698071d", "component_budget": { "node_count": 2, "item_count": 2, @@ -22908,18 +22877,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are unrelated: one requests a new token-classification model class for Qwen3_5, while the other asks for FP8 training support. They should not be clustered as duplicates.", + "summary": "These two issues are unrelated: one requests a new Qwen3_5 token-classification model class, while the other asks for FP8 training support. They should not be clustered as duplicates.", "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a duplicate of the other; they address different areas (model head support vs. training precision). There is no clear canonical issue in this pair.", + "best_issue_reason": "Issue 46295 is the stronger standalone tracker in this pair because it has broader scope and more inbound attention, but it is not a duplicate of 45810.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45810", "right": "issue:46295", "accept": false, - "reason": "Different requests with no shared code path or concrete bug: adding Qwen3_5ForTokenClassification versus FP8 training support." + "reason": "Different underlying requests and code paths: model-class support for token classification vs training precision/FP8 support." } ] }, @@ -22959,88 +22928,88 @@ "issue:45871|issue:45873", "issue:45872|issue:45873" ], - "prepared_review_unit_hash": "cd1683c049ed2fa277d7c84f073f38989f23fd1a9ed57a8424cdff838f8e42f9", + "prepared_review_unit_hash": "8e013e0a5b70691c570b48ca71e33f00ae88ff6b547342bffef112dbcfd60e42", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5719, - "estimated_input_tokens": 1430, - "estimated_eval_tokens": 3116 + "serialized_chars": 5726, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5719, - "estimated_input_tokens": 1430, - "estimated_eval_tokens": 3116 + "serialized_chars": 5726, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The set contains two separate exact-duplicate issue pairs: 45870/45871 and 45872/45873. Issue 45869 is a different financial dataset request and should not be merged with either pair.", - "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue fits the whole set because it splits into two unrelated duplicate topics; within the 'Financial time series prediction with DataForge' pair, 45870 is the cleanest representative.", + "summary": "This cluster contains two exact duplicate pairs and three distinct financial/DataForge-related issues. Accept the two exact-duplicate pairs (45870/45871 and 45872/45873); reject the cross-pair similarity matches and the dataset-vs-example match.", + "confidence": 0.96, + "canonical_issue_reason": "issue:45870 is the best canonical issue for its duplicate pair because 45871 is identical, and 45870 is the earlier report with a clear, specific title.", "canonical_pr_reason": null, - "best_issue_reason": "45870 is the best overall representative because it is an exact duplicate of 45871, appears first, and has the clearest focused example title among the items.", + "best_issue_reason": "issue:45870 is the strongest issue representative in this cluster because it is an exact duplicate target with the cleanest, earliest metadata; the other issues are thematically related but not the same underlying request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45869", "right": "issue:45870", "accept": false, - "reason": "Same finance domain, but 45869 is a dataset bundle promo/request while 45870 is an example/tutorial; not the same underlying issue." + "reason": "Different artifacts: a dataset bundle offer vs a DataForge prediction example. Same financial theme, but not the same issue." }, { "left": "issue:45869", "right": "issue:45871", "accept": false, - "reason": "Different artifact types: dataset bundle versus example request. Shared theme is too broad to treat as a duplicate." + "reason": "Different artifacts: a dataset bundle offer vs a DataForge prediction example. Shared topic is too broad to treat as duplicates." }, { "left": "issue:45870", "right": "issue:45871", "accept": true, - "reason": "Same title and same body size, with no distinguishing signals; this looks like a direct duplicate." + "reason": "Exact duplicate pair: identical title and matching metadata/body length, so they appear to describe the same underlying issue." }, { "left": "issue:45870", "right": "issue:45872", "accept": false, - "reason": "Both mention financial time series, but 45872 adds a scraping service component; these are different feature requests." + "reason": "Related DataForge financial content, but one is a prediction example and the other is a scraping-service request; not the same underlying issue." }, { "left": "issue:45870", "right": "issue:45873", "accept": false, - "reason": "Different scope: example/prediction request versus example plus scraping service. Not the same concrete change." + "reason": "Related DataForge financial content, but one is a prediction example and the other is a scraping-service request; not the same underlying issue." }, { "left": "issue:45871", "right": "issue:45872", "accept": false, - "reason": "Exact duplicate of the example issue on one side, but the other side is a different 'scraping service' request." + "reason": "Related financial/DataForge topic, but these describe different concrete requests: prediction example vs scraping service." }, { "left": "issue:45871", "right": "issue:45873", "accept": false, - "reason": "Same mismatch as above: example/prediction request versus example plus scraping service." + "reason": "Related financial/DataForge topic, but these describe different concrete requests: prediction example vs scraping service." }, { "left": "issue:45872", "right": "issue:45873", "accept": true, - "reason": "Identical title and body size strongly indicate a direct duplicate." + "reason": "Exact duplicate pair: identical title and matching metadata/body length, so they appear to describe the same underlying issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: 45870/45871 and 45872/45873 are well-supported duplicate pairs from identical titles and body lengths, while 45869 is meaningfully different by title and should stay separate. The soft-edge rejections are conservative, and the canonical-issue framing matches the observed split." + "feedback": "The cluster summary is grounded in the packet: 45870/45871 and 45872/45873 are the only pairs with identical titles and equal body lengths, while the remaining cross-pair matches are only broad soft similarities and are conservatively rejected. The canonical choice of the earlier report in each duplicate pair is also reasonable." }, "evaluator_used": true, "retried": false, @@ -23067,40 +23036,40 @@ "soft_pairs": [ "issue:46439|issue:46441" ], - "prepared_review_unit_hash": "4f23a3258c7311f4ce6383688472df19e6ee8b26dd5e3541d76e37c21ba532e6", + "prepared_review_unit_hash": "54a71e62c711c52c0a1ef241f1d7049bc492260affb82df3347e253e3f45ec33", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1640, - "estimated_input_tokens": 410, - "estimated_eval_tokens": 1076 + "serialized_chars": 1641, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1640, - "estimated_input_tokens": 410, - "estimated_eval_tokens": 1076 + "serialized_chars": 1641, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both items concern static-cache failures around `linear_attention`, but they describe different concrete breakages: one is an early-initialization shape corruption leading to `RuntimeError`, the other is a Qwen3.5 `KeyError` for a missing cache entry. They are related subsystem reports, not clearly the same bug.", - "confidence": 0.67, - "canonical_issue_reason": "Issue 46439 is the better canonical issue because it states the broader underlying problem in static-cache `early_initialization` affecting linear-attention layers, rather than a model-specific symptom.", + "summary": "Both issues involve `cache_implementation=\"static\"` and linear-attention/Qwen3.5 failures, but they report different concrete breakages: one is an `early_initialization` shape corruption leading to `RuntimeError`, the other is a missing `linear_attention` cache path causing `KeyError`. Related area, but not the same underlying bug.", + "confidence": 0.78, + "canonical_issue_reason": "Issue 46439 is the better canonical issue because it names the more specific mechanism (`early_initialization`) and the observed failure mode, making it a stronger anchor for the static-cache/linear-attention bug family.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46439 is the best representative of the cluster since it names the likely root cause and general failure mode, making it more useful for grouping related reports.", + "best_issue_reason": "Issue 46439 is the best representative of the cluster because it describes the broader root-cause-style failure, while 46441 looks more like a specific downstream symptom for a particular model/configuration.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46439", "right": "issue:46441", "accept": false, - "reason": "Related area, but not clearly the same underlying bug: one reports shape corruption during early initialization, the other a `KeyError` for `linear_attention` in Qwen3.5. Different failure signatures and likely different fixes." + "reason": "Do not merge: both touch static cache and linear attention, but the reported problems differ materially (`RuntimeError` from shape corruption vs `KeyError` for missing `linear_attention`). They are related symptoms, not clearly the same defect." } ] }, @@ -23130,7 +23099,7 @@ "soft_pairs": [ "issue:46551|issue:46552" ], - "prepared_review_unit_hash": "fda07d6cdf51a143c0ca43dda418b66deb1a60465c60737837d67e2a8834d1c6", + "prepared_review_unit_hash": "90927896cee9f5f7dc604b7b3b5cec7fb0dd1a8150dc5d519a23a627d018c754", "component_budget": { "node_count": 2, "item_count": 2, @@ -23152,27 +23121,27 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are near-perfect duplicates: same title, same body, same tokenizer regression in v5. Issue #46551 is the best canonical/best issue choice because it was created slightly earlier and is otherwise identical.", - "confidence": 0.99, - "canonical_issue_reason": "Issue #46551 is the earliest of two identical reports and cleanly represents the duplicate cluster.", + "summary": "The two closed issues appear to be duplicates: they have identical titles, the same body length, and strong soft-similarity around the same Wav2Vec2CTCTokenizer v5 bug affecting get_special_tokens_mask and convert_ids_to_tokens. The packet does not prove the body text is exactly identical, but it is still strong duplicate evidence.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #46552 is the better canonical target because it reports the same bug as #46551 and has slightly higher discussion/inbound reference activity, making it the more central duplicate thread.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #46551 is the strongest issue representative because it is identical to the other report and posted first.", + "best_issue_reason": "Issue #46552 is the best representative for the cluster because it fully captures the underlying tokenizer bug and has the stronger conversation/reference signal.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46551", "right": "issue:46552", "accept": true, - "reason": "Same title and same body describing the same Wav2Vec2CTCTokenizer v5 special-token leak bug; this is a clear duplicate pair." + "reason": "Accept: identical titles, equal body length, and explicit soft-similarity indicate the same underlying issue report, even though exact body text identity is not proven." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The assessment is grounded in the packet: the two issues have identical titles, identical body length, and the soft-similarity signal is maximal (Jaccard 1.0). Choosing #46551 as canonical because it was created earlier is conservative and well supported. No PR-pair concerns apply here." + "accept": false, + "feedback": "The duplicate conclusion is supported, but the summary overstates the evidence by saying the issues have identical bodies. The packet directly shows identical titles, equal body length, and strong soft-similarity, but not the exact body text. The canonical/best issue choice is otherwise reasonable given the higher discussion and inbound reference activity on #46552." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -23196,12 +23165,12 @@ "soft_pairs": [ "issue:46657|issue:46692" ], - "prepared_review_unit_hash": "d1517f0c19fd926881a14467fd3e4f28ea7f5a5bd6765735680603af731e6fa3", + "prepared_review_unit_hash": "5ad904b4ded6bb59f98ffb382a2902d3dee2f20ffdd08438530ae47e673b0ad3", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1569, + "serialized_chars": 1570, "estimated_input_tokens": 393, "estimated_eval_tokens": 1042 }, @@ -23209,7 +23178,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1569, + "serialized_chars": 1570, "estimated_input_tokens": 393, "estimated_eval_tokens": 1042 }, @@ -23218,18 +23187,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one reports a tensor size mismatch during object-detection training, the other reports out-of-memory after several epochs while training RT-DETR v2. The overlap is only broad training context.", + "summary": "These two issues are only loosely related by being training-time problems in detection models, but they describe different failure modes: a tensor size mismatch versus an out-of-memory condition after several epochs. They do not look like duplicates.", "confidence": 0.95, - "canonical_issue_reason": "Issue #46657 is the more specific, actionable report, but there is no true canonical issue because the two tickets describe different failure modes and likely different root causes.", + "canonical_issue_reason": "Issue #46657 is the better canonical issue because it reports a concrete, specific runtime error during object detection training, making it the clearer representative bug report.", "canonical_pr_reason": null, - "best_issue_reason": "#46657 is the best representative issue for the cluster only because it is the clearer, more concrete bug report; #46692 is a separate OOM symptom and should remain distinct.", + "best_issue_reason": "Issue #46657 is the best issue overall because it is more specific and actionable; #46692 appears to be a different symptom and likely a different underlying problem.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46657", "right": "issue:46692", "accept": false, - "reason": "Reject: one issue is a tensor-shape mismatch, the other is an OOM after epochs. Same broad training area, but not the same underlying bug." + "reason": "Reject: the reported failures are different in kind and likely in root cause. #46657 is a tensor shape/size mismatch, while #46692 is an out-of-memory issue after some epochs during RT-DETR v2 training. Same broad training area, but not the same bug." } ] }, @@ -23259,12 +23228,12 @@ "soft_pairs": [ "issue:46822|issue:46823" ], - "prepared_review_unit_hash": "0a6a350a6649aab44c069dbe227b5d28487c53e97581251133ea65c81e3001a7", + "prepared_review_unit_hash": "363c1012dd0fe4dff99ea18c9847c533e833f7660c7272e19152d112f3343f37", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1674, + "serialized_chars": 1673, "estimated_input_tokens": 419, "estimated_eval_tokens": 1094 }, @@ -23272,7 +23241,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1674, + "serialized_chars": 1673, "estimated_input_tokens": 419, "estimated_eval_tokens": 1094 }, @@ -23281,18 +23250,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues describe different bugs: one is a ZeRO-3 zero.Init partitioning/OOM problem, the other is a 4-bit device_map/max_memory dispatch-to-CPU error. They should not be clustered as duplicates.", - "confidence": 0.98, - "canonical_issue_reason": null, + "summary": "These are two separate model-loading bugs: one about ZeRO-3 zero.Init failing to partition a composite multimodal submodule and causing OOM, and one about 4-bit auto device mapping sending a model to CPU despite sufficient memory. They do not look like duplicates or the same underlying fix.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 46822 is the more concrete, unresolved bug report with a specific ZeRO-3 partitioning failure and clear OOM symptom.", "canonical_pr_reason": null, - "best_issue_reason": "No single canonical issue fits both reports because they hit different code paths and failure modes; 46823 is more concrete and already closed, but it is not a duplicate of 46822.", + "best_issue_reason": "Issue 46822 is the best representative issue in this cluster because it has a precise failure mode and is still open.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46822", "right": "issue:46823", "accept": false, - "reason": "Different underlying bugs: ZeRO-3 composite submodule partitioning vs 4-bit auto device mapping with max_memory causing CPU dispatch/ValueError. Shared theme is model loading, but not the same defect." + "reason": "Different subsystems and failure modes: ZeRO-3 partitioning/OOM vs 4-bit device_map auto-placement/CPU dispatch ValueError. They are not the same underlying bug." } ] }, @@ -23308,13 +23277,13 @@ "component_index": 25, "component_count": 28, "review_unit_index": 1, - "review_unit_count": 1, + "review_unit_count": 2, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-46906-22", + "cluster_id": "cluster-46906-23", "nodes": [ "issue:46906", "issue:46940", @@ -23337,7 +23306,8 @@ "issue:47377", "issue:47390", "issue:47406", - "issue:47419" + "issue:47419", + "issue:47442" ], "soft_pairs": [ "issue:46906|issue:46940", @@ -23346,526 +23316,592 @@ "issue:46906|issue:46985", "issue:46906|issue:47037", "issue:46940|issue:46948", - "issue:46940|issue:46969", - "issue:46940|issue:46985", - "issue:46940|issue:46998", "issue:46948|issue:46969", "issue:46948|issue:46985", + "issue:46940|issue:46969", + "issue:46940|issue:46985", + "issue:47037|issue:47049", "issue:46948|issue:46998", + "issue:47049|issue:47053", + "issue:46998|issue:47049", "issue:46969|issue:46985", - "issue:46969|issue:47037", - "issue:46969|issue:47091", - "issue:46985|issue:47037", + "issue:46940|issue:46998", "issue:46985|issue:47049", - "issue:46998|issue:47053", + "issue:47049|issue:47091", "issue:46998|issue:47037", - "issue:46998|issue:47049", - "issue:46985|issue:46998", - "issue:46998|issue:47091", - "issue:47037|issue:47049", + "issue:46985|issue:47037", + "issue:46969|issue:47037", "issue:47037|issue:47091", "issue:47037|issue:47053", - "issue:47049|issue:47053", - "issue:47049|issue:47091", + "issue:46969|issue:47091", "issue:47053|issue:47091", - "issue:47053|issue:47102", "issue:47091|issue:47102", - "issue:46985|issue:47091", + "issue:46998|issue:47053", + "issue:47053|issue:47102", + "issue:46985|issue:46998", + "issue:46998|issue:47091", "issue:47037|issue:47102", + "issue:46985|issue:47091", "issue:46985|issue:47102", "issue:47049|issue:47102", - "issue:47120|issue:47182", + "issue:47265|issue:47279", + "issue:47265|issue:47309", + "issue:47279|issue:47309", + "issue:47265|issue:47377", + "issue:47279|issue:47377", + "issue:47256|issue:47265", + "issue:47256|issue:47279", + "issue:47182|issue:47256", "issue:47120|issue:47256", + "issue:47309|issue:47377", + "issue:47102|issue:47256", + "issue:47182|issue:47265", + "issue:47120|issue:47182", "issue:47120|issue:47279", "issue:47120|issue:47265", "issue:47120|issue:47377", "issue:47182|issue:47377", - "issue:47182|issue:47256", - "issue:47182|issue:47265", "issue:47182|issue:47279", "issue:47227|issue:47377", "issue:47227|issue:47256", + "issue:47182|issue:47309", + "issue:47309|issue:47419", "issue:47227|issue:47279", "issue:47182|issue:47227", "issue:47227|issue:47265", - "issue:47256|issue:47265", - "issue:47256|issue:47279", - "issue:47102|issue:47256", - "issue:47265|issue:47279", - "issue:47265|issue:47309", - "issue:47265|issue:47377", - "issue:47279|issue:47309", - "issue:47279|issue:47377", - "issue:47309|issue:47377", - "issue:47182|issue:47309", - "issue:47309|issue:47419", - "issue:47390|issue:47406", - "issue:47390|issue:47419", - "issue:47377|issue:47390", - "issue:47227|issue:47390", - "issue:47279|issue:47390", - "issue:47406|issue:47419", - "issue:47377|issue:47406", - "issue:47279|issue:47406", - "issue:47120|issue:47406", + "issue:47419|issue:47442", "issue:47377|issue:47419", + "issue:47377|issue:47442", "issue:47279|issue:47419", - "issue:47265|issue:47419" + "issue:47406|issue:47419", + "issue:47279|issue:47442", + "issue:47227|issue:47442", + "issue:47390|issue:47419", + "issue:47390|issue:47406", + "issue:47390|issue:47442", + "issue:47406|issue:47442", + "issue:47377|issue:47406", + "issue:47279|issue:47406" ], - "prepared_review_unit_hash": "37ff8569a64aa3b3480988269ceb6a0fbaf7eb8b13c67954b9dbf14a86d866ab", + "prepared_review_unit_hash": "9c801eb099da9facd78e6b528c9faf210229be344cd8e81e82ef3d9a80c41e0b", "component_budget": { - "node_count": 22, - "item_count": 22, - "soft_pair_count": 71, - "serialized_chars": 37832, - "estimated_input_tokens": 9458, - "estimated_eval_tokens": 19172 + "node_count": 23, + "item_count": 23, + "soft_pair_count": 74, + "serialized_chars": 39466, + "estimated_input_tokens": 9867, + "estimated_eval_tokens": 19990 }, "budget": { - "node_count": 22, - "item_count": 22, - "soft_pair_count": 71, - "serialized_chars": 37832, - "estimated_input_tokens": 9458, - "estimated_eval_tokens": 19172 + "node_count": 23, + "item_count": 23, + "soft_pair_count": 72, + "serialized_chars": 34132, + "estimated_input_tokens": 8533, + "estimated_eval_tokens": 17322 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, - "split": false, + "split": true, "analyst_result": { - "summary": "All items are date-stamped \"[serge] integration failure triage\" issues. The similarity is from a shared template and recurring triage workflow, not from a single underlying bug. I do not see a true duplicate cluster here; these are separate daily reports.", - "confidence": 0.97, - "canonical_issue_reason": "No true canonical duplicate: each issue is a separate daily serge integration triage report for a different date, so they should not be merged as the same bug.", + "summary": "This cluster is a date-stamped series of [serge] integration failure triage issues. They look like recurring daily reports, not duplicate incidents, so the soft similarity links should not be merged.", + "confidence": 0.92, + "canonical_issue_reason": "If a single representative is needed, issue 47442 is the strongest anchor: it is the most recent open triage report and has strong inbound reference activity. But the cluster itself is a recurring series, not one duplicate bug.", "canonical_pr_reason": null, - "best_issue_reason": "If one issue must be kept as the active tracker, issue:47419 is the newest open report; however, it is still a date-specific triage issue, not a duplicate of the others.", + "best_issue_reason": "Issue 47442 is the best overall representative because it is current, open, and appears to be the active landing point for later references.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46906", "right": "issue:46940", "accept": false, - "reason": "Same template/title, but different daily triage snapshots for different dates; no evidence of the same underlying failure." + "reason": "Same template and subsystem, but different dates imply separate triage reports rather than the same underlying failure." }, { "left": "issue:46906", "right": "issue:46948", "accept": false, - "reason": "Date-separated triage reports; similarity is template-level only, not a shared bug." + "reason": "These are adjacent daily triage issues, not evidence of one identical bug instance." }, { "left": "issue:46906", "right": "issue:46969", "accept": false, - "reason": "Different day\u2019s triage report; not the same concrete issue." + "reason": "Different report dates; this looks like recurring monitoring output, not a duplicate issue." }, { "left": "issue:46906", "right": "issue:46985", "accept": false, - "reason": "Recurring tracker format, but these are distinct dated reports." + "reason": "Same recurring triage series, but no sign of the same concrete incident." }, { "left": "issue:46906", "right": "issue:47037", "accept": false, - "reason": "Same triage series, but separate daily snapshots rather than one duplicate bug." + "reason": "Separated by several days and likely different failures; too weak for duplicate merge." }, { "left": "issue:46940", "right": "issue:46948", "accept": false, - "reason": "Different dates in the same triage series; no shared concrete failure identified." + "reason": "Highly similar titles, but they are date-specific triage entries for different days." }, { - "left": "issue:46940", + "left": "issue:46948", "right": "issue:46969", "accept": false, - "reason": "Template match only; these appear to be separate daily reports." + "reason": "Same recurring triage pattern, but not the same underlying bug report." }, { - "left": "issue:46940", + "left": "issue:46948", "right": "issue:46985", "accept": false, - "reason": "Different integration triage dates, so not the same underlying issue." + "reason": "Different dates and likely different integration runs; not a duplicate." }, { "left": "issue:46940", - "right": "issue:46998", + "right": "issue:46969", "accept": false, - "reason": "Same broad workflow, but not the same bug or change." + "reason": "Recurring report series only; no concrete shared incident identified." }, { - "left": "issue:46948", - "right": "issue:46969", + "left": "issue:46940", + "right": "issue:46985", "accept": false, - "reason": "Distinct dated triage issues; similarity is superficial." + "reason": "Same title family, but different day-specific triage issues." }, { - "left": "issue:46948", - "right": "issue:46985", + "left": "issue:47037", + "right": "issue:47049", "accept": false, - "reason": "Separate daily reports, not a duplicate bug." + "reason": "Adjacent daily reports, but still separate triage artifacts." }, { "left": "issue:46948", "right": "issue:46998", "accept": false, - "reason": "No evidence they describe the same failure, only the same triage template." + "reason": "The similarity is template-level; the dated reports are distinct." }, { - "left": "issue:46969", - "right": "issue:46985", + "left": "issue:47049", + "right": "issue:47053", "accept": false, - "reason": "Consecutive daily triage reports; likely different snapshots/incidents." + "reason": "Same recurring triage series, but no evidence they describe one identical failure." }, { - "left": "issue:46969", - "right": "issue:47037", + "left": "issue:46998", + "right": "issue:47049", "accept": false, - "reason": "Different date and report content scope; not the same underlying problem." + "reason": "Different dates and separate integration triage entries." }, { "left": "issue:46969", - "right": "issue:47091", + "right": "issue:46985", "accept": false, - "reason": "Same series, but each issue is a separate daily triage entry." + "reason": "Date-stamped daily triage reports; not the same bug." }, { - "left": "issue:46985", - "right": "issue:47037", + "left": "issue:46940", + "right": "issue:46998", "accept": false, - "reason": "Different daily triage reports rather than one duplicated issue." + "reason": "Same broad topic, but these are separate daily triage issues." }, { "left": "issue:46985", "right": "issue:47049", "accept": false, - "reason": "Template-only similarity; no shared concrete code-path bug." + "reason": "Recurring triage template only; no shared concrete failure shown." }, { - "left": "issue:46998", - "right": "issue:47053", + "left": "issue:47049", + "right": "issue:47091", "accept": false, - "reason": "Separate dated triage entries, not the same bug." + "reason": "Similar report family, but different incident dates." }, { "left": "issue:46998", "right": "issue:47037", "accept": false, - "reason": "Same triage tracker format, but different daily incidents." + "reason": "Daily triage issues should not be merged on title similarity alone." }, { - "left": "issue:46998", - "right": "issue:47049", + "left": "issue:46985", + "right": "issue:47037", "accept": false, - "reason": "Different dates and likely different failure snapshots." + "reason": "Different dates and likely different integration failures." }, { - "left": "issue:46985", - "right": "issue:46998", + "left": "issue:46969", + "right": "issue:47037", "accept": false, - "reason": "Adjacent daily reports; not enough to treat as duplicates." + "reason": "Looks like the same recurring report template, not a duplicate bug." }, { - "left": "issue:46998", + "left": "issue:47037", "right": "issue:47091", "accept": false, - "reason": "Recurring triage series only; no concrete shared failure shown." + "reason": "Separate daily triage artifacts, not one underlying issue." }, { "left": "issue:47037", - "right": "issue:47049", + "right": "issue:47053", "accept": false, - "reason": "Different day\u2019s triage report; not the same underlying issue." + "reason": "Same recurring integration triage series; no concrete duplicate signal." }, { - "left": "issue:47037", + "left": "issue:46969", "right": "issue:47091", "accept": false, - "reason": "Same broad integration-triage theme, but separate reports." + "reason": "Different report dates; insufficient evidence of one merged incident." }, { - "left": "issue:47037", - "right": "issue:47053", + "left": "issue:47053", + "right": "issue:47091", "accept": false, - "reason": "No explicit shared target or code-path; these are likely distinct daily snapshots." + "reason": "Similar subsystem and title, but not the same dated report." }, { - "left": "issue:47049", - "right": "issue:47053", - "accept": false, - "reason": "Different dates in the same tracking series; not a duplicate." - }, - { - "left": "issue:47049", - "right": "issue:47091", + "left": "issue:47091", + "right": "issue:47102", "accept": false, - "reason": "Same template, but separate triage instances." + "reason": "Adjacent daily triage issues, but still distinct reports." }, { - "left": "issue:47053", - "right": "issue:47091", + "left": "issue:46998", + "right": "issue:47053", "accept": false, - "reason": "Date-stamped reports, not one underlying bug." + "reason": "Template similarity only; these are separate daily triage entries." }, { "left": "issue:47053", "right": "issue:47102", "accept": false, - "reason": "Different daily integration triage issues; no mergeable duplicate evidence." + "reason": "Different days in the same triage series, not the same bug." }, { - "left": "issue:47091", - "right": "issue:47102", + "left": "issue:46985", + "right": "issue:46998", "accept": false, - "reason": "Separate triage days; similarity is only in the title and structure." + "reason": "Same recurring report pattern, but not a duplicate incident." }, { - "left": "issue:46985", + "left": "issue:46998", "right": "issue:47091", "accept": false, - "reason": "Recurring report format, but not the same concrete failure." + "reason": "No evidence they refer to one specific failure; just repeated triage." }, { - "left": "issue:47037", - "right": "issue:47102", + "left": "issue:46985", + "right": "issue:47091", "accept": false, - "reason": "Different dated snapshots in the serge triage series." + "reason": "Date-specific reports from the same series should remain separate." }, { "left": "issue:46985", "right": "issue:47102", "accept": false, - "reason": "Same umbrella topic, but each issue is a distinct daily report." + "reason": "Recurring triage issues, but not the same underlying problem." }, { "left": "issue:47049", "right": "issue:47102", "accept": false, - "reason": "No sign of a shared bug; these are separate day-by-day triage entries." + "reason": "Similar text, different daily report context." }, { - "left": "issue:47120", - "right": "issue:47182", + "left": "issue:47265", + "right": "issue:47279", "accept": false, - "reason": "Same triage template, but different dates and likely different failure sets." + "reason": "These are sequential daily triage issues, not one duplicated bug report." }, { - "left": "issue:47120", - "right": "issue:47256", + "left": "issue:47265", + "right": "issue:47309", "accept": false, - "reason": "Separate daily integration-failure triage reports; not the same issue." + "reason": "Same report template, but distinct dated incidents." }, { - "left": "issue:47120", - "right": "issue:47279", + "left": "issue:47279", + "right": "issue:47309", "accept": false, - "reason": "Different dated snapshots; similarity is superficial." + "reason": "High text similarity is just the daily triage format; do not merge." }, { - "left": "issue:47120", - "right": "issue:47265", + "left": "issue:47265", + "right": "issue:47377", "accept": false, - "reason": "Recurring triage tracker, not a duplicate bug report." + "reason": "Different dates and likely different failure snapshots." }, { - "left": "issue:47120", + "left": "issue:47279", "right": "issue:47377", "accept": false, - "reason": "Same series, but these are distinct daily reports." + "reason": "Same recurring integration triage series, but not the same concrete issue." }, { - "left": "issue:47182", - "right": "issue:47377", + "left": "issue:47256", + "right": "issue:47265", + "accept": false, + "reason": "Adjacent daily reports, but no shared underlying bug is shown." + }, + { + "left": "issue:47256", + "right": "issue:47279", "accept": false, - "reason": "Different day\u2019s triage entry; no shared concrete fix target." + "reason": "These are separate day-by-day triage artifacts." }, { "left": "issue:47182", "right": "issue:47256", "accept": false, - "reason": "Separate daily reports rather than duplicate issues." + "reason": "Same family of reports, but different dates and likely different incidents." }, { - "left": "issue:47182", - "right": "issue:47265", + "left": "issue:47120", + "right": "issue:47256", "accept": false, - "reason": "Same broad topic, but distinct triage snapshots." + "reason": "Recurring triage entry, not a duplicate fix target." }, { - "left": "issue:47182", - "right": "issue:47279", + "left": "issue:47102", + "right": "issue:47256", "accept": false, - "reason": "No evidence of the same underlying bug; only shared template text." + "reason": "Too far apart in date and context to be the same underlying bug." }, { - "left": "issue:47227", - "right": "issue:47377", + "left": "issue:47182", + "right": "issue:47265", "accept": false, - "reason": "Date-separated triage reports, not one duplicate problem." + "reason": "Both are triage reports, but separate daily instances." }, { - "left": "issue:47227", - "right": "issue:47256", + "left": "issue:47120", + "right": "issue:47182", "accept": false, - "reason": "Same tracking pattern, but different report dates." + "reason": "The similarity is only the shared reporting template." }, { - "left": "issue:47227", + "left": "issue:47120", "right": "issue:47279", "accept": false, - "reason": "Distinct daily integration-failure triage entries." + "reason": "Different days in a recurring triage series." }, { - "left": "issue:47182", - "right": "issue:47227", + "left": "issue:47120", + "right": "issue:47265", "accept": false, - "reason": "Adjacent reports in the series, but still separate daily snapshots." + "reason": "Same subsystem and title, but not the same incident." }, { - "left": "issue:47227", - "right": "issue:47265", + "left": "issue:47120", + "right": "issue:47377", "accept": false, - "reason": "Template overlap only; no shared concrete code-path issue." + "reason": "Recurring daily triage issue, not a duplicate bug." }, { - "left": "issue:47256", - "right": "issue:47265", + "left": "issue:47182", + "right": "issue:47377", "accept": false, - "reason": "Different daily triage reports with the same naming convention." + "reason": "Date-stamped reports should remain distinct unless they name the same failure." }, { - "left": "issue:47256", + "left": "issue:47182", "right": "issue:47279", "accept": false, - "reason": "Not the same bug; just recurring serge triage reporting." + "reason": "Same report pattern, but separate triage entries." }, { - "left": "issue:47102", - "right": "issue:47256", + "left": "issue:47227", + "right": "issue:47377", "accept": false, - "reason": "Separate dated issue reports, not duplicates." + "reason": "No evidence these point to one concrete code-path problem." }, { - "left": "issue:47265", - "right": "issue:47279", + "left": "issue:47227", + "right": "issue:47256", "accept": false, - "reason": "Same triage series, but different days and likely different failures." + "reason": "Adjacent daily reports, but not duplicates." }, { - "left": "issue:47265", + "left": "issue:47182", "right": "issue:47309", "accept": false, - "reason": "Distinct daily triage entries; no same underlying issue shown." + "reason": "Same recurring integration triage format; different dates." }, { - "left": "issue:47265", + "left": "issue:47309", "right": "issue:47377", "accept": false, - "reason": "Different dates and report instances; not a duplicate." + "reason": "These are sequential daily reports, not one merged failure." }, { - "left": "issue:47279", - "right": "issue:47309", + "left": "issue:47182", + "right": "issue:47227", "accept": false, - "reason": "Recurring template only; these are separate reports." + "reason": "Only title/template similarity; different daily incidents." }, { - "left": "issue:47279", - "right": "issue:47377", + "left": "issue:47227", + "right": "issue:47279", "accept": false, - "reason": "No shared concrete bug or fix target, just similar triage text." + "reason": "Recurring triage series, but not the same bug." }, { - "left": "issue:47309", - "right": "issue:47377", + "left": "issue:47227", + "right": "issue:47265", "accept": false, - "reason": "Different daily integration-triage snapshots." + "reason": "Date-specific reports with no shared concrete failure evidence." }, { - "left": "issue:47182", - "right": "issue:47309", + "left": "issue:47419", + "right": "issue:47442", "accept": false, - "reason": "Same broad serge triage topic, but not the same failure." + "reason": "Same series, but the open report is a later day and should not be merged as a duplicate." }, { - "left": "issue:47309", + "left": "issue:47377", "right": "issue:47419", "accept": false, - "reason": "Separate day-stamped reports; not a duplicate issue." + "reason": "Recurring daily triage issues, not the same underlying incident." }, { - "left": "issue:47390", - "right": "issue:47406", + "left": "issue:47377", + "right": "issue:47442", "accept": false, - "reason": "Same triage series, but different dates and likely different incidents." + "reason": "Different report dates; title similarity alone is insufficient." }, { - "left": "issue:47390", + "left": "issue:47279", "right": "issue:47419", "accept": false, - "reason": "Different daily tracker entries; no evidence they describe one bug." + "reason": "Separate daily triage artifacts from the same reporting stream." }, { - "left": "issue:47377", - "right": "issue:47390", + "left": "issue:47406", + "right": "issue:47419", "accept": false, - "reason": "Adjacent daily triage issues, but not the same underlying problem." + "reason": "Same triage template, but distinct day-specific issues." }, { - "left": "issue:47227", - "right": "issue:47390", + "left": "issue:47279", + "right": "issue:47442", "accept": false, - "reason": "Shared title pattern only; distinct dated reports." + "reason": "No indication these are the same concrete bug or change." }, { - "left": "issue:47279", - "right": "issue:47390", + "left": "issue:47227", + "right": "issue:47442", "accept": false, - "reason": "Recurring tracker format, but separate failure snapshots." + "reason": "Later report in the same series, but not a duplicate of the earlier one." }, { - "left": "issue:47406", + "left": "issue:47390", "right": "issue:47419", "accept": false, - "reason": "Different dates in the same triage stream; not a duplicate." + "reason": "Recurring integration triage entries; different dates and likely different failures." }, { - "left": "issue:47377", + "left": "issue:47390", "right": "issue:47406", "accept": false, - "reason": "Separate daily integration triage issues." + "reason": "High similarity comes from the common triage template, not a shared bug." }, { - "left": "issue:47279", - "right": "issue:47406", + "left": "issue:47390", + "right": "issue:47442", "accept": false, - "reason": "No indication of the same concrete bug; only title similarity." + "reason": "Distinct daily reports; do not collapse them into one issue." }, { - "left": "issue:47120", - "right": "issue:47406", + "left": "issue:47406", + "right": "issue:47442", "accept": false, - "reason": "Same reporting template, but different daily entries." + "reason": "Same recurring triage series, but separate report instances." }, { "left": "issue:47377", - "right": "issue:47419", + "right": "issue:47406", "accept": false, - "reason": "Different date-stamped triage reports; not the same issue." + "reason": "Different dates in the same recurring triage stream." }, { "left": "issue:47279", - "right": "issue:47419", + "right": "issue:47406", + "accept": false, + "reason": "Template similarity only; no concrete duplicate evidence." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 25, + "component_count": 28, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46906-23", + "nodes": [ + "issue:47227", + "issue:47377", + "issue:47390" + ], + "soft_pairs": [ + "issue:47377|issue:47390", + "issue:47227|issue:47390" + ], + "prepared_review_unit_hash": "f9f22b33a22607112efa27ee138c67acb3c96440ed100fcb1347fc572d15cfe2", + "component_budget": { + "node_count": 23, + "item_count": 23, + "soft_pair_count": 74, + "serialized_chars": 39466, + "estimated_input_tokens": 9867, + "estimated_eval_tokens": 19990 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2318, + "estimated_input_tokens": 580, + "estimated_eval_tokens": 1416 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These look like recurring daily triage issues for serge integration failures, not the same underlying bug. They share a template-like title but have different dates and no linking evidence.", + "confidence": 0.76, + "canonical_issue_reason": "Issue #47377 is the best representative because it has the highest inbound reference count and matches the recurring triage pattern.", + "canonical_pr_reason": null, + "best_issue_reason": "#47377 is the strongest issue to keep as the cluster representative; the set appears to be multiple dated triage reports rather than duplicates of one defect.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:47377", + "right": "issue:47390", "accept": false, - "reason": "Recurring serge integration triage reports, but distinct instances." + "reason": "Same triage template, but the dated titles indicate separate daily reports with no evidence of a single shared failure." }, { - "left": "issue:47265", - "right": "issue:47419", + "left": "issue:47227", + "right": "issue:47390", "accept": false, - "reason": "Same high-level triage theme, but not a duplicate bug report." + "reason": "Highly similar wording, but these are date-specific triage issues and not clearly the same underlying bug or change." } ] }, @@ -23895,7 +23931,7 @@ "soft_pairs": [ "issue:46983|issue:46984" ], - "prepared_review_unit_hash": "08b45285c8e87c765842aa7a9d7be07d72a926ca60b320101e2c6bef1981b1f3", + "prepared_review_unit_hash": "dd0d5445b1451e992dea6e9d12c26018b3c612d4cb70f30963dff3aabb9b1563", "component_budget": { "node_count": 2, "item_count": 2, @@ -23917,28 +23953,28 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are near-certain duplicates: same title, same feature request, and highly similar bodies. Keep one as the canonical representative and close the other as duplicate.", - "confidence": 0.98, - "canonical_issue_reason": "issue #46984 is the better representative because it has some discussion activity, suggesting it accumulated more context, while otherwise matching #46983 almost exactly.", + "summary": "The two closed issues are highly similar requests for RWKV7 model support, likely describing the same feature request, but the evidence is only soft similarity and does not justify a definitive duplicate call. No PRs are present in the cluster.", + "confidence": 0.88, + "canonical_issue_reason": "#46984 is the better canonical issue because it has the same title as #46983 and slightly more discussion activity, making it the more established representative thread.", "canonical_pr_reason": null, - "best_issue_reason": "issue #46984 is the best issue in the cluster for the same reason: it appears to be the more active duplicate and is equally on-topic for RWKV7 support.", + "best_issue_reason": "#46984 is the strongest representative for the cluster since it matches the shared RWKV7 support request exactly and has a bit more activity.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46983", "right": "issue:46984", - "accept": true, - "reason": "Same exact title and clearly the same underlying feature request (RWKV7 model support); this is a duplicate pair." + "accept": false, + "reason": "The issues are very similar and appear to request the same RWKV7 support, but the packet only provides soft-similarity evidence and deterministic_accept is false, so this should be treated as likely related rather than a definitive duplicate." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The duplicate call is grounded by the identical title, very high body similarity, and the modest activity difference supports picking #46984 as the representative. The wording is a bit strong, but not beyond the packet evidence." + "accept": false, + "feedback": "The cluster summary is mostly grounded, but the verdict is a bit too strong. The packet only provides soft-similarity evidence between the two issues, and deterministic_accept is false, so calling them definitive duplicates overstates the evidence. It would be safer to say they are highly similar / likely the same request for RWKV7 support. The canonical issue choice is reasonable based on slightly higher discussion activity, and the note that there are no PRs is supported." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -23961,12 +23997,12 @@ "soft_pairs": [ "issue:47007|issue:47015" ], - "prepared_review_unit_hash": "9365bc277847bd7a09da27648dbe83ebc43d69c4e2e98fc5e71eea286ac8de5e", + "prepared_review_unit_hash": "1022a6665bf08a9a4f44330a44d99756406279f02c764c91bb4f6866dc70a13a", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1625, + "serialized_chars": 1626, "estimated_input_tokens": 407, "estimated_eval_tokens": 1070 }, @@ -23974,7 +24010,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1625, + "serialized_chars": 1626, "estimated_input_tokens": 407, "estimated_eval_tokens": 1070 }, @@ -23983,18 +24019,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are thematically similar AI/cybersecurity writeups, but they do not look like the same underlying report or change. This is a soft topical overlap, not a duplicate.", - "confidence": 0.84, - "canonical_issue_reason": "Issue 47015 is the broader, more umbrella-like AI-vulnerability-discovery topic, but neither issue is a clear duplicate of the other.", + "summary": "Both issues are broad, essay-like posts about AI-driven vulnerability discovery, but they do not describe a concrete shared bug or code change. They are thematically related rather than clear duplicates, so the soft pair should be rejected.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 47015 is the more general, umbrella-style title and reads like the cleaner representative of the topic cluster.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 47015 is the better representative of the cluster because it is broader and more canonical in framing; 47007 reads more like a specific lessons-learned variant.", + "best_issue_reason": "Issue 47015 is the best representative because it is broader and more directly names the underlying subject matter without the extra 'lessons from using advanced LLMs' framing.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:47007", "right": "issue:47015", "accept": false, - "reason": "They share the same general AI vulnerability-discovery theme, but the titles indicate different angles and likely different content rather than one underlying issue." + "reason": "The titles overlap on AI-based vulnerability discovery, but there is no concrete bug, code path, or explicit shared change. These look like related high-level writeups, not the same underlying issue." } ] }, @@ -24024,7 +24060,7 @@ "soft_pairs": [ "issue:47138|issue:47139" ], - "prepared_review_unit_hash": "b3d53ebb9d62ea3f0c532adbf0c97d5fef516af307099f10784648dd2b47e9d6", + "prepared_review_unit_hash": "16883a79be63f75afdb2e5e8500d5b1b44df2c5d04bc4c09e79d439424a74ec3", "component_budget": { "node_count": 2, "item_count": 2, @@ -24046,24 +24082,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are duplicate reports of the exact same bug report/title and should be clustered together as one issue.", + "summary": "The two items are effectively duplicate issues with identical titles and near-identical bodies about the same /v1 base_url routing bug in transformers chat management endpoints.", "confidence": 0.99, - "canonical_issue_reason": "Issue #47138 is the better canonical representative because it was created slightly earlier and has more inbound references, while the titles/bodies are effectively identical.", + "canonical_issue_reason": "Issue 47138 is the better canonical issue because it was created slightly earlier and has more inbound references, while describing the same problem as 47139.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #47138 is the best issue to keep as the primary tracker for this duplicate pair for the same reasons: earliest creation and marginally higher reference activity.", + "best_issue_reason": "47138 is the strongest single issue representative of this bug due to earliest creation time and slightly higher discussion/reference activity.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:47138", "right": "issue:47139", "accept": true, - "reason": "Same title, same reported behavior, and no distinguishing evidence; these are the same underlying issue." + "reason": "Same title, same apparent body length/content, and the same underlying routing bug; these are duplicates of one issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The clustering decision is grounded: both issues have the same title, identical body length, and a soft-similarity score of 1.0, with only a small difference in inbound references and creation time. The choice of issue:47138 as the canonical tracker is conservative and supported by the packet." + "feedback": "The analysis is grounded in the packet: both issues have identical titles, identical body length, and a soft-similarity score of 1.0, with only a small difference in creation time and inbound references. The duplicate verdict and canonical choice are conservative enough for the available evidence." }, "evaluator_used": true, "retried": false, @@ -24082,7 +24118,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-40546-81", + "cluster_id": "cluster-40546-82", "nodes": [ "pull_request:43251", "pull_request:43254", @@ -24096,12 +24132,10 @@ "pull_request:46912", "pull_request:46957", "pull_request:46958", - "pull_request:46974", "pull_request:46995", "pull_request:47003", "pull_request:47077", "pull_request:47078", - "pull_request:47082", "pull_request:47090", "pull_request:47093", "pull_request:47105", @@ -24122,16 +24156,18 @@ "pull_request:47284", "pull_request:47291", "pull_request:47292", - "pull_request:47295", "pull_request:47301", - "pull_request:47304", "pull_request:47336", + "pull_request:47357", "pull_request:47360", "pull_request:47374", "pull_request:47383", "pull_request:47384", "pull_request:47395", - "pull_request:47420" + "pull_request:47420", + "pull_request:47422", + "pull_request:47446", + "pull_request:47447" ], "soft_pairs": [ "pull_request:43251|pull_request:43254", @@ -24140,34 +24176,37 @@ "pull_request:46903|pull_request:46958", "pull_request:46903|pull_request:47395", "pull_request:47268|pull_request:47284", - "pull_request:46958|pull_request:46974", "pull_request:47173|pull_request:47284", "pull_request:47202|pull_request:47204", "pull_request:47204|pull_request:47374", "pull_request:47284|pull_request:47292", "pull_request:47108|pull_request:47202", "pull_request:47077|pull_request:47078", + "pull_request:47204|pull_request:47243", "pull_request:47202|pull_request:47360", "pull_request:47108|pull_request:47360", - "pull_request:47204|pull_request:47243", "pull_request:47105|pull_request:47129", - "pull_request:46266|pull_request:46957", "pull_request:47077|pull_request:47212", + "pull_request:46266|pull_request:46957", "pull_request:47301|pull_request:47360", + "pull_request:47360|pull_request:47447", "pull_request:47360|pull_request:47383", - "pull_request:47216|pull_request:47360", + "pull_request:47384|pull_request:47447", "pull_request:44775|pull_request:47275", "pull_request:47166|pull_request:47173", "pull_request:47291|pull_request:47384", "pull_request:47077|pull_request:47275", + "pull_request:47336|pull_request:47446", "pull_request:47275|pull_request:47301", + "pull_request:47422|pull_request:47446", "pull_request:47216|pull_request:47275", - "pull_request:47163|pull_request:47284", "pull_request:46912|pull_request:47284", + "pull_request:47163|pull_request:47284", "pull_request:46196|pull_request:47202", "pull_request:46912|pull_request:47179", - "pull_request:46836|pull_request:47202", + "pull_request:47204|pull_request:47357", "pull_request:46196|pull_request:47210", + "pull_request:46836|pull_request:47202", "pull_request:47003|pull_request:47204", "pull_request:47093|pull_request:47275", "pull_request:45157|pull_request:47093", @@ -24175,312 +24214,330 @@ "pull_request:47163|pull_request:47173", "pull_request:47108|pull_request:47420", "pull_request:45157|pull_request:46395", - "pull_request:46836|pull_request:47275", - "pull_request:46912|pull_request:47082", - "pull_request:47204|pull_request:47295", - "pull_request:47163|pull_request:47384", - "pull_request:44775|pull_request:47301", - "pull_request:46196|pull_request:46995", - "pull_request:47301|pull_request:47384", - "pull_request:47216|pull_request:47384", - "pull_request:47243|pull_request:47304", - "pull_request:47336|pull_request:47384", - "pull_request:47077|pull_request:47301", - "pull_request:47166|pull_request:47292", - "pull_request:44775|pull_request:47077" + "pull_request:46836|pull_request:47275" ], - "prepared_review_unit_hash": "017b689b71ec4cd958a3986d19ec6749561f951429c0a3afc5b89d816ac1d745", + "prepared_review_unit_hash": "443f7bb7a59e01c9385e3c858c203d0d89d3c3072b4513b98c4c3bab5d827a39", "component_budget": { - "node_count": 81, - "item_count": 81, - "soft_pair_count": 225, - "serialized_chars": 279853, - "estimated_input_tokens": 69964, - "estimated_eval_tokens": 140184 + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 54, - "serialized_chars": 127466, - "estimated_input_tokens": 31867, - "estimated_eval_tokens": 63990 + "soft_pair_count": 45, + "serialized_chars": 125413, + "estimated_input_tokens": 31354, + "estimated_eval_tokens": 62964 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated PRs; only the fixed_cross_entropy kwargs pair and the CohereASR double-shift fixes look like true duplicates. The rest are separate docs, model-addition, test, or infrastructure changes.", - "confidence": 0.78, + "summary": "This is a heterogeneous cluster with several unrelated PRs. Only a few edges look like true duplicates: the fixed_cross_entropy kwargs fixes, and the CohereASR/PPFormulaNet training-loss double-shift fixes. Most other pairs are same broad area or just docs/tests, not the same underlying change.", + "confidence": 0.47, "canonical_issue_reason": null, - "canonical_pr_reason": "No issue artifacts are present; among the PRs, pull_request:47090 is the best canonical representative for the training-loss double-shift fix family because it targets the same concrete bug across multiple models, has an explicit issue link, and is the broadest/most complete version of that fix.", + "canonical_pr_reason": "If forced to pick one PR representative, 47090 is the strongest canonical candidate among the genuine duplicates: it addresses the same concrete double-shift loss bug, covers the affected ASR-related models more broadly, and has the clearest issue linkage/tests.", "best_issue_reason": null, - "best_pr_reason": "pull_request:47090 is the strongest representative of the duplicate set around the double-shift loss bug: it addresses the same code-path problem in several affected models and is more complete than the narrower CohereASR-only variants.", + "best_pr_reason": "47090 is the best standalone PR in the cluster because it is a concrete code fix for a real bug, not just docs/tests, and it is the most complete version of the actual duplicate group here.", "soft_edge_verdicts": [ { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Same helper, same explicit issue, same kwargs passthrough fix in loss_utils.py." + "reason": "Same underlying cross-entropy kwargs bug in the same helper; both patch the same code path and could be merged into one PR." }, { "left": "pull_request:46903", "right": "pull_request:47090", "accept": true, - "reason": "Same underlying double-shift training-loss bug pattern; 47090 is a broader version of the same fix across overlapping model code paths." + "reason": "Same training-loss double-shift bug in the same model family; 47090 is the broader version of the same fix." }, { "left": "pull_request:47090", "right": "pull_request:47395", "accept": false, - "reason": "Different model and different loss path; same general symptom, not the same concrete change." + "reason": "Different models and different loss paths; both are double-shift fixes, but not the same code-path problem." }, { "left": "pull_request:46903", "right": "pull_request:46958", "accept": true, - "reason": "Both fix the same CohereASR/PPFormulaNet training-loss double-shift bug; implementation differs but the underlying defect is the same." + "reason": "Same CohereASR/PPFormulaNet double-shift loss bug; these are alternative implementations of the same fix." }, { "left": "pull_request:46903", "right": "pull_request:47395", "accept": false, - "reason": "Different models and code paths; only the bug theme is similar." + "reason": "Different model family and unrelated loss logic; shared theme is too broad to treat as duplicate." }, { "left": "pull_request:47268", "right": "pull_request:47284", "accept": false, - "reason": "Separate test fixes for unrelated models." - }, - { - "left": "pull_request:46958", - "right": "pull_request:46974", - "accept": false, - "reason": "One is a loss bug fix, the other is a generation typo fix in Musicgen." + "reason": "Unrelated failing tests in different model families; no shared underlying bug." }, { "left": "pull_request:47173", "right": "pull_request:47284", "accept": false, - "reason": "Unrelated XPU test updates for different models." + "reason": "Both are test-only changes, but for different models and different failures." }, { "left": "pull_request:47202", "right": "pull_request:47204", "accept": false, - "reason": "Both touch CI/docker setup, but one deletes a workflow and the other adds exporter deps; not the same change." + "reason": "Both touch Docker/CI, but one drops a builder and the other adds exporter dependencies; not the same change." }, { "left": "pull_request:47204", "right": "pull_request:47374", "accept": false, - "reason": "Exporter test/runtime setup versus docs-only exporter documentation." + "reason": "Executorch Docker test deps versus exporter docs; related ecosystem, but different artifacts and goals." }, { "left": "pull_request:47284", "right": "pull_request:47292", "accept": false, - "reason": "Unrelated model test fixes." + "reason": "Different model test fixes with no shared code path or bug." }, { "left": "pull_request:47108", "right": "pull_request:47202", "accept": false, - "reason": "Dependabot comment cleanup versus docker-image-builder removal." + "reason": "Doc-builder pin comment cleanup is unrelated to removing the CPU/light docker builder." }, { "left": "pull_request:47077", "right": "pull_request:47078", "accept": false, - "reason": "Overlapping Japanese docs edits, but one is a translation PR and the other is wording polish; not the same underlying change." + "reason": "Both are Japanese doc edits, but one is a translation PR and the other is wording cleanup; not the same underlying change." + }, + { + "left": "pull_request:47204", + "right": "pull_request:47243", + "accept": false, + "reason": "Exporter Docker image dependencies versus exporter code fixes; related area, but not one mergeable bug fix." }, { "left": "pull_request:47202", "right": "pull_request:47360", "accept": false, - "reason": "Different CI/workflow topics." + "reason": "Different CI workflow changes with no common underlying fix." }, { "left": "pull_request:47108", "right": "pull_request:47360", "accept": false, - "reason": "Different workflow changes with no shared concrete bug." + "reason": "Unrelated documentation comment tweak versus OTEL CI workflow change." }, { - "left": "pull_request:47204", - "right": "pull_request:47243", + "left": "pull_request:47105", + "right": "pull_request:47129", "accept": false, - "reason": "Exporter dependency setup versus exporter implementation fixes; related area but distinct changes." + "reason": "VITS test re-enable versus RT-DETR converter fix; unrelated model areas." }, { - "left": "pull_request:47105", - "right": "pull_request:47129", + "left": "pull_request:47077", + "right": "pull_request:47212", "accept": false, - "reason": "Separate model/test issues." + "reason": "Japanese translation edits versus kv-cache docs updates; same language only, not same change." }, { "left": "pull_request:46266", "right": "pull_request:46957", "accept": false, - "reason": "Different model-family modernization versus RoPE test work." + "reason": "Different subsystems: checkpoint conversion mapping/model support versus generation kernel optimization." }, { - "left": "pull_request:47077", - "right": "pull_request:47212", + "left": "pull_request:47301", + "right": "pull_request:47360", "accept": false, - "reason": "Different docs topics." + "reason": "Assisted-decoding docs versus CI observability workflow; no shared bug or change." }, { - "left": "pull_request:47301", - "right": "pull_request:47360", + "left": "pull_request:47360", + "right": "pull_request:47447", "accept": false, - "reason": "Docs for MTP versus CI workflow OTEL pilot; unrelated." + "reason": "CI workflow changes versus speculative-decoding cache rollback; unrelated." }, { "left": "pull_request:47360", "right": "pull_request:47383", "accept": false, - "reason": "Different areas: CI workflow versus Kimi integration tests." + "reason": "CI workflow change versus Kimi integration tests; not the same artifact." }, { - "left": "pull_request:47216", - "right": "pull_request:47360", + "left": "pull_request:47384", + "right": "pull_request:47447", "accept": false, - "reason": "Quantization docs/code versus CI workflow changes." + "reason": "Gemma 4 heterogeneous config change versus cache rollback feature; different code paths." }, { "left": "pull_request:44775", "right": "pull_request:47275", "accept": false, - "reason": "Both are docs/model additions, but they are unrelated features." + "reason": "Both are docs/model additions, but for different subjects and no shared underlying change." }, { "left": "pull_request:47166", "right": "pull_request:47173", "accept": false, - "reason": "Separate quantization and XPU test updates." + "reason": "Different test suites and failures; no evidence of the same bug." }, { "left": "pull_request:47291", "right": "pull_request:47384", "accept": false, - "reason": "Generation-config deprecation versus heterogeneous per-layer config support; related subsystem, different change." + "reason": "Both touch generation/configuration, but one deprecates a config path and the other changes heterogeneous per-layer config." }, { "left": "pull_request:47077", "right": "pull_request:47275", "accept": false, - "reason": "Different docs/model work." + "reason": "Japanese docs translation versus adding a new model; unrelated." }, { - "left": "pull_request:47216", - "right": "pull_request:47275", + "left": "pull_request:47336", + "right": "pull_request:47446", "accept": false, - "reason": "Compressed-tensors work versus model addition; unrelated." + "reason": "Offloading save bugfix versus typo/comment cleanup; not duplicate changes." }, { - "left": "pull_request:47163", - "right": "pull_request:47284", + "left": "pull_request:47275", + "right": "pull_request:47301", + "accept": false, + "reason": "New model support versus assisted-decoding docs; unrelated." + }, + { + "left": "pull_request:47422", + "right": "pull_request:47446", + "accept": false, + "reason": "CI/docker pinning fix versus comment typo cleanup; different scope and code path." + }, + { + "left": "pull_request:47216", + "right": "pull_request:47275", "accept": false, - "reason": "Paged-attention XPU tests versus Mimo_v2_flash tests." + "reason": "Quantization integration work versus a new vision model; unrelated." }, { "left": "pull_request:46912", "right": "pull_request:47284", "accept": false, - "reason": "RoPE test updates versus Mimo_v2_flash test fixes." + "reason": "RoPE test improvements versus MIMO v2 flash test fixes; unrelated." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47284", + "accept": false, + "reason": "Different XPU-related test changes for different components; not the same bug." }, { "left": "pull_request:46196", "right": "pull_request:47202", "accept": false, - "reason": "SDPA dispatch support versus docker image builder cleanup." + "reason": "SDPA dispatch model support versus docker builder removal; unrelated." }, { "left": "pull_request:46912", - "right": "pull_request:47082", + "right": "pull_request:47179", + "accept": false, + "reason": "RoPE test changes versus Granite SWA model support; different goals and code paths." + }, + { + "left": "pull_request:47204", + "right": "pull_request:47357", + "accept": false, + "reason": "Exporter test dependencies versus FSDP CI/save changes; not the same underlying issue." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47210", + "accept": false, + "reason": "Both are attention-related, but one adds SDPA dispatch and the other routes packed rows through varlen attention; different implementation problems." + }, + { + "left": "pull_request:46836", + "right": "pull_request:47202", "accept": false, - "reason": "RoPE tests versus a new keypoint-matcher model." + "reason": "Unlimited OCR model support versus docker builder removal; unrelated." }, { "left": "pull_request:47003", "right": "pull_request:47204", "accept": false, - "reason": "OpenVINO exporter implementation versus exporter test-image dependencies; distinct PRs." + "reason": "OpenVINO exporter work versus ExecuTorch/torch-light dependency changes; different exporter targets." }, { "left": "pull_request:47093", "right": "pull_request:47275", "accept": false, - "reason": "Different new-model additions." + "reason": "OpenPangu model addition versus EfficientViT-SAM model addition; same broad model-support theme, but distinct architectures and PRs." }, { "left": "pull_request:45157", "right": "pull_request:47093", "accept": false, - "reason": "Different model support efforts." + "reason": "Prism GGUF support versus new model addition; unrelated." }, { - "left": "pull_request:46836", - "right": "pull_request:47275", + "left": "pull_request:46995", + "right": "pull_request:47163", "accept": false, - "reason": "Different model additions and code paths." + "reason": "Availability check for flash linear attention versus paged-attention tests; different layers of the stack." }, { - "left": "pull_request:46912", - "right": "pull_request:47284", + "left": "pull_request:47163", + "right": "pull_request:47173", "accept": false, - "reason": "Separate test changes, not the same bug." + "reason": "Both are XPU-related tests, but they target different failures and code paths." }, { - "left": "pull_request:46196", - "right": "pull_request:47210", + "left": "pull_request:47108", + "right": "pull_request:47420", "accept": false, - "reason": "SDPA model dispatch support versus packed-SDPA varlen attention routing." + "reason": "Doc-builder comment formatting versus tokenization special-token return cleanup; unrelated." }, { - "left": "pull_request:47003", - "right": "pull_request:47204", + "left": "pull_request:45157", + "right": "pull_request:46395", "accept": false, - "reason": "Exporter implementation versus exporter test/runtime setup; not the same concrete fix." + "reason": "Prism GGUF dequantization support versus Param2MoE model support; different features." }, { - "left": "pull_request:47093", + "left": "pull_request:46836", "right": "pull_request:47275", "accept": false, - "reason": "Unrelated model additions." - }, - { - "left": "pull_request:44775", - "right": "pull_request:47301", - "accept": false, - "reason": "Docs-only changes in different topic areas." + "reason": "Unlimited OCR support versus EfficientViT-SAM support; unrelated model additions." }, { - "left": "pull_request:47166", - "right": "pull_request:47292", + "left": "pull_request:46957", + "right": "pull_request:47210", "accept": false, - "reason": "Different XPU-related test updates for different models." + "reason": "Finegrained FP8 decode optimization versus packed SDPA varlen attention; different generation optimizations." }, { - "left": "pull_request:47291", - "right": "pull_request:47384", + "left": "pull_request:47275", + "right": "pull_request:47301", "accept": false, - "reason": "Generation configuration deprecation versus heterogeneous config validation." + "reason": "Model addition versus MTP documentation; no shared underlying change." }, { - "left": "pull_request:47077", - "right": "pull_request:47301", + "left": "pull_request:47422", + "right": "pull_request:47446", "accept": false, - "reason": "Japanese docs wording/translation versus assisted-decoding MTP docs." + "reason": "Docker/CI pinning and RNNT fix versus typo-only comment cleanup; not duplicates." }, { - "left": "pull_request:44775", - "right": "pull_request:47077", + "left": "pull_request:47384", + "right": "pull_request:47447", "accept": false, - "reason": "Different docs changes with no shared underlying feature or bug." + "reason": "Heterogeneous Gemma config change versus speculative decoding cache rollback; different code paths and features." } ] }, @@ -24502,7 +24559,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-40546-81", + "cluster_id": "cluster-40546-82", "nodes": [ "pull_request:44517", "pull_request:44775", @@ -24514,15 +24571,14 @@ "pull_request:46604", "pull_request:46658", "pull_request:46836", - "pull_request:46895", "pull_request:46912", "pull_request:46957", "pull_request:46995", "pull_request:47003", + "pull_request:47077", "pull_request:47078", "pull_request:47082", "pull_request:47093", - "pull_request:47105", "pull_request:47108", "pull_request:47113", "pull_request:47143", @@ -24538,39 +24594,61 @@ "pull_request:47275", "pull_request:47292", "pull_request:47294", - "pull_request:47298", + "pull_request:47295", "pull_request:47301", "pull_request:47304", "pull_request:47306", "pull_request:47320", "pull_request:47322", "pull_request:47336", - "pull_request:47352", - "pull_request:47360", + "pull_request:47357", "pull_request:47367", "pull_request:47383", + "pull_request:47384", "pull_request:47420", "pull_request:47422", - "pull_request:47438" + "pull_request:47446", + "pull_request:47447" ], "soft_pairs": [ - "pull_request:47367|pull_request:47420", + "pull_request:47204|pull_request:47295", + "pull_request:46912|pull_request:47082", + "pull_request:47163|pull_request:47384", + "pull_request:44775|pull_request:47301", + "pull_request:47301|pull_request:47447", + "pull_request:46196|pull_request:46995", + "pull_request:47301|pull_request:47384", + "pull_request:47216|pull_request:47384", + "pull_request:47243|pull_request:47304", + "pull_request:47292|pull_request:47446", + "pull_request:47322|pull_request:47446", + "pull_request:47077|pull_request:47301", + "pull_request:47243|pull_request:47447", + "pull_request:46604|pull_request:47357", + "pull_request:44775|pull_request:47077", + "pull_request:47166|pull_request:47292", "pull_request:47294|pull_request:47320", + "pull_request:47166|pull_request:47446", + "pull_request:47367|pull_request:47420", "pull_request:46267|pull_request:47383", "pull_request:47113|pull_request:47422", "pull_request:46912|pull_request:47173", - "pull_request:46266|pull_request:47301", - "pull_request:47093|pull_request:47143", + "pull_request:47163|pull_request:47357", "pull_request:46509|pull_request:46995", + "pull_request:47093|pull_request:47143", "pull_request:46995|pull_request:47143", - "pull_request:46266|pull_request:47383", + "pull_request:47357|pull_request:47383", + "pull_request:46266|pull_request:47301", "pull_request:46196|pull_request:47179", - "pull_request:47082|pull_request:47210", + "pull_request:46266|pull_request:47383", "pull_request:44775|pull_request:47304", "pull_request:47093|pull_request:47163", "pull_request:47179|pull_request:47210", - "pull_request:47082|pull_request:47108", + "pull_request:47082|pull_request:47210", + "pull_request:47294|pull_request:47447", + "pull_request:46509|pull_request:47357", "pull_request:46604|pull_request:47003", + "pull_request:47082|pull_request:47108", "pull_request:47108|pull_request:47301", "pull_request:46266|pull_request:47216", "pull_request:47078|pull_request:47275", @@ -24581,392 +24659,313 @@ "pull_request:44775|pull_request:46836", "pull_request:47163|pull_request:47179", "pull_request:46604|pull_request:46658", - "pull_request:47082|pull_request:47216", + "pull_request:47082|pull_request:47383", "pull_request:44517|pull_request:45157", - "pull_request:47275|pull_request:47336", "pull_request:47210|pull_request:47304", + "pull_request:47275|pull_request:47336", "pull_request:46957|pull_request:47216", "pull_request:47216|pull_request:47336", "pull_request:47306|pull_request:47420", - "pull_request:47360|pull_request:47422", - "pull_request:47204|pull_request:47422", - "pull_request:46912|pull_request:47078", - "pull_request:47322|pull_request:47420", - "pull_request:47179|pull_request:47301", - "pull_request:47166|pull_request:47422", - "pull_request:47166|pull_request:47352", - "pull_request:47298|pull_request:47438", - "pull_request:47275|pull_request:47422", - "pull_request:47298|pull_request:47420", - "pull_request:47143|pull_request:47320", - "pull_request:46912|pull_request:47294", - "pull_request:47175|pull_request:47294", - "pull_request:47175|pull_request:47438", - "pull_request:46895|pull_request:47306", - "pull_request:46196|pull_request:47301", - "pull_request:47243|pull_request:47301", - "pull_request:46912|pull_request:47438", - "pull_request:47105|pull_request:47298", - "pull_request:47292|pull_request:47383", - "pull_request:45157|pull_request:46658", - "pull_request:47078|pull_request:47322", - "pull_request:44517|pull_request:47301", - "pull_request:47294|pull_request:47438", - "pull_request:47113|pull_request:47304", - "pull_request:47294|pull_request:47360" + "pull_request:47204|pull_request:47422" ], - "prepared_review_unit_hash": "76c4699b4648b2ad4e39579fbde1ed3cbe5cacb4de45bbb46adc850cdad1c390", + "prepared_review_unit_hash": "eb67cf1c58525597716028d5cb5a640f694d072aeb96275ab8674cd47f68f9ea", "component_budget": { - "node_count": 81, - "item_count": 81, - "soft_pair_count": 225, - "serialized_chars": 279853, - "estimated_input_tokens": 69964, - "estimated_eval_tokens": 140184 + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 60, - "serialized_chars": 132552, - "estimated_input_tokens": 33138, - "estimated_eval_tokens": 66532 + "soft_pair_count": 56, + "serialized_chars": 133672, + "estimated_input_tokens": 33418, + "estimated_eval_tokens": 67092 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a heterogeneous collection of unrelated PRs (new model additions, docs, CI, exporter/quantization changes, and isolated bug fixes). I found no true duplicate pair among the soft edges.", - "confidence": 0.99, + "summary": "This cluster is highly heterogeneous: it mixes model additions, documentation updates, CI/test tweaks, quantization/export changes, and a few generation/runtime fixes. The soft links are superficial topical similarity rather than the same underlying bug or change, so none should be merged as duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the cluster is too diverse and the soft edges do not indicate the same underlying change or bug.", + "canonical_pr_reason": "No single canonical PR: the items are not a duplicate set and cover unrelated changes across docs, models, tests, CI, quantization, and generation.", "best_issue_reason": null, - "best_pr_reason": "No single PR is a good global representative; the items span unrelated subsystems and cannot be merged into one canonical duplicate target.", + "best_pr_reason": "No global best PR; the cluster is too heterogeneous to safely pick one representative as the canonical duplicate target.", "soft_edge_verdicts": [ { - "left": "pull_request:47367", - "right": "pull_request:47420", + "left": "pull_request:47204", + "right": "pull_request:47295", "accept": false, - "reason": "Docs/pipeline device-placement clarification is unrelated to a tokenizer special-token return cleanup." + "reason": "Docker/export-test dependency changes vs a DiffusionGemma mask fix; unrelated code paths." }, { - "left": "pull_request:47294", - "right": "pull_request:47320", + "left": "pull_request:46912", + "right": "pull_request:47082", "accept": false, - "reason": "Kernel-native causal_conv1d utilities and response_schema deprecation address different subsystems and different changes." + "reason": "RoPE test data updates vs a new RoMa model implementation; not the same change." }, { - "left": "pull_request:46267", - "right": "pull_request:47383", + "left": "pull_request:47163", + "right": "pull_request:47384", "accept": false, - "reason": "Serve chat parsing refactor is unrelated to Kimi integration tests." + "reason": "Paged-attention XPU tests vs Gemma 4 heterogeneous config changes; different subsystems." }, { - "left": "pull_request:47113", - "right": "pull_request:47422", + "left": "pull_request:44775", + "right": "pull_request:47301", "accept": false, - "reason": "Audio feature/proc backend requirements are not the same as CI torchaudio pinning and RNN-T fix work." + "reason": "Parallelism docs vs assisted-decoding MTP docs; distinct documentation topics." }, { - "left": "pull_request:46912", - "right": "pull_request:47173", + "left": "pull_request:47301", + "right": "pull_request:47447", "accept": false, - "reason": "RoPE test updates and XPU expectation fixes are separate test maintenance changes, not the same bug." + "reason": "MTP docs vs speculative-decoding cache rollback; different features and code paths." }, { - "left": "pull_request:46266", - "right": "pull_request:47301", + "left": "pull_request:46196", + "right": "pull_request:46995", "accept": false, - "reason": "Dino modernization/model conversion work is unrelated to multi-token prediction docs/features." + "reason": "SDPA dispatch support vs flash-linear-attention availability; related attention theme but different backend support." }, { - "left": "pull_request:47093", - "right": "pull_request:47143", + "left": "pull_request:47301", + "right": "pull_request:47384", "accept": false, - "reason": "OpenPangu-V2 model support and NeuCodec support are distinct model additions." + "reason": "MTP docs vs Gemma 4 config refactor; unrelated changes." }, { - "left": "pull_request:46509", - "right": "pull_request:46995", + "left": "pull_request:47216", + "right": "pull_request:47384", "accept": false, - "reason": "Nemotron Omni support and flash-linear-attention MLU availability are unrelated changes." + "reason": "Compressed-tensors FP8 support vs Gemma 4 config changes; not the same bug or feature." }, { - "left": "pull_request:46995", - "right": "pull_request:47143", + "left": "pull_request:47243", + "right": "pull_request:47304", "accept": false, - "reason": "Backend availability detection and NeuCodec model support do not share an underlying code path." + "reason": "ExecuTorch exporter fixes vs an Apple MPS docs cleanup; unrelated." }, { - "left": "pull_request:46266", - "right": "pull_request:47383", + "left": "pull_request:47292", + "right": "pull_request:47446", "accept": false, - "reason": "Dino modernization/model conversion is unrelated to Kimi integration tests." + "reason": "XPU test expectation updates vs typo fixes in comments; different scopes entirely." }, { - "left": "pull_request:46196", - "right": "pull_request:47179", + "left": "pull_request:47322", + "right": "pull_request:47446", "accept": false, - "reason": "SDPA dispatch support for wav2vec2-family models is unrelated to Granite-SWA model support." + "reason": "Quark config API change vs typo-only cleanup; not duplicates." }, { - "left": "pull_request:47082", - "right": "pull_request:47210", + "left": "pull_request:47077", + "right": "pull_request:47301", "accept": false, - "reason": "RoMa model support and packed-SDPA varlen attention routing are different features." + "reason": "Japanese DistilBERT docs vs MTP docs; unrelated documentation." }, { - "left": "pull_request:44775", - "right": "pull_request:47304", + "left": "pull_request:47243", + "right": "pull_request:47447", "accept": false, - "reason": "N-D parallelism docs and MPS graph-cache docs are unrelated documentation edits." + "reason": "ExecuTorch exporter fixes vs speculative-decoding cache rollback; different runtime areas." }, { - "left": "pull_request:47093", - "right": "pull_request:47163", + "left": "pull_request:46604", + "right": "pull_request:47357", "accept": false, - "reason": "OpenPangu-V2 support and XPU paged-attention tests are unrelated." + "reason": "Mistral tekken tokenizer utilities vs FSDP CI/save changes; unrelated functionality." }, { - "left": "pull_request:47179", - "right": "pull_request:47210", + "left": "pull_request:44775", + "right": "pull_request:47077", "accept": false, - "reason": "Granite-SWA model support is not the same as packed SDPA varlen attention optimization." + "reason": "Parallelism docs vs Japanese DistilBERT docs; no shared underlying change." }, { - "left": "pull_request:47082", - "right": "pull_request:47108", + "left": "pull_request:47166", + "right": "pull_request:47292", "accept": false, - "reason": "RoMa model support and a doc-builder comment tweak are unrelated." + "reason": "GPTQ test updates vs TIPSv2 DPT XPU test fixes; different model/test targets." }, { - "left": "pull_request:46604", - "right": "pull_request:47003", + "left": "pull_request:47294", + "right": "pull_request:47320", "accept": false, - "reason": "Mistral tekken tokenizer utilities and OpenVINO exporter support are unrelated efforts." + "reason": "Kernel-native causal_conv1d work vs response_schema deprecation; unrelated changes." }, { - "left": "pull_request:47108", - "right": "pull_request:47301", + "left": "pull_request:47166", + "right": "pull_request:47446", "accept": false, - "reason": "Dependabot-compatible workflow comment changes are unrelated to MTP docs." + "reason": "GPTQ test changes vs comment typo cleanup; not the same issue." }, { - "left": "pull_request:46266", - "right": "pull_request:47216", + "left": "pull_request:47367", + "right": "pull_request:47420", "accept": false, - "reason": "Dino model conversion work and compressed-tensors fp8 support are not the same change." + "reason": "Pipeline device-placement docs vs tokenizer special-token builder cleanup; unrelated." }, { - "left": "pull_request:47078", - "right": "pull_request:47275", + "left": "pull_request:46267", + "right": "pull_request:47383", "accept": false, - "reason": "Japanese docs wording cleanup and EfficientViT-SAM model support are unrelated." + "reason": "Serve chat parsing update vs Kimi integration tests; different areas of the stack." }, { - "left": "pull_request:47082", - "right": "pull_request:47275", + "left": "pull_request:47113", + "right": "pull_request:47422", "accept": false, - "reason": "RoMa keypoint matcher support and EfficientViT-SAM support are separate model additions." + "reason": "Audio feature/proc requirement fixes vs Docker torch/torchaudio pinning; related domain but different problems." }, { - "left": "pull_request:46266", - "right": "pull_request:47336", + "left": "pull_request:47093", + "right": "pull_request:47143", "accept": false, - "reason": "Dino modernization and fully offloaded model saving are different bugs/features." + "reason": "OpenPangu-V2 model support vs NeuCodec model support; separate model additions." }, { "left": "pull_request:46995", - "right": "pull_request:47301", + "right": "pull_request:47143", "accept": false, - "reason": "Flash-linear-attention availability and MTP documentation are unrelated." + "reason": "Flash-linear-attention availability on MLU vs NeuCodec model support; unrelated." }, { - "left": "pull_request:47175", - "right": "pull_request:47179", + "left": "pull_request:47357", + "right": "pull_request:47383", "accept": false, - "reason": "Olmo3 CI expectation updates and Granite-SWA model support do not describe the same underlying change." + "reason": "FSDP CI/save work vs Kimi integration tests; no shared code-path fix." }, { - "left": "pull_request:44775", - "right": "pull_request:46836", + "left": "pull_request:46266", + "right": "pull_request:47301", "accept": false, - "reason": "Parallelism docs and Unlimited OCR model support are unrelated." + "reason": "Dino modernization work vs MTP docs; completely different topics." }, { - "left": "pull_request:47163", + "left": "pull_request:46196", "right": "pull_request:47179", "accept": false, - "reason": "XPU paged-attention test additions and Granite-SWA support are unrelated." + "reason": "SDPA dispatch support vs Granite-swa model support; unrelated." }, { - "left": "pull_request:46604", - "right": "pull_request:46658", + "left": "pull_request:46266", + "right": "pull_request:47383", "accept": false, - "reason": "Mistral tokenizer utilities and Step-3.7 Flash model-building/conversion work are different changes." + "reason": "Dino modernization work vs Kimi integration tests; not duplicates." }, { - "left": "pull_request:47082", - "right": "pull_request:47216", + "left": "pull_request:44775", + "right": "pull_request:47304", "accept": false, - "reason": "RoMa model support and compressed-tensors fp8 support are unrelated." + "reason": "Parallelism docs vs MPS graph-cache docs; different documentation updates." }, { - "left": "pull_request:44517", - "right": "pull_request:45157", + "left": "pull_request:47093", + "right": "pull_request:47163", "accept": false, - "reason": "Qwen3-TTS model support and PrismML Bonsai support are unrelated model additions." + "reason": "OpenPangu-V2 model support vs XPU paged-attention tests; unrelated." }, { - "left": "pull_request:47275", - "right": "pull_request:47336", + "left": "pull_request:47179", + "right": "pull_request:47210", "accept": false, - "reason": "EfficientViT-SAM support and offloading save bugfix are unrelated." + "reason": "Granite-swa model support vs packed SDPA varlen routing; different changes." }, { - "left": "pull_request:47210", - "right": "pull_request:47304", + "left": "pull_request:47082", + "right": "pull_request:47210", "accept": false, - "reason": "Packed SDPA varlen attention routing and MPS graph-cache docs are unrelated." + "reason": "RoMa model addition vs SDPA varlen packing; unrelated." }, { - "left": "pull_request:46957", - "right": "pull_request:47216", + "left": "pull_request:47294", + "right": "pull_request:47447", "accept": false, - "reason": "Finegrained FP8 decode optimization and compressed-tensors fp8 support are different quantization efforts." + "reason": "causal_conv1d kernel integration vs speculative-decoding cache rollback; different functionality." }, { - "left": "pull_request:47216", - "right": "pull_request:47336", + "left": "pull_request:46509", + "right": "pull_request:47357", "accept": false, - "reason": "Compressed-tensors fp8 work and fully offloaded model saving are unrelated." + "reason": "Nemotron Omni model support vs FSDP CI/save changes; no shared bug or change." }, { - "left": "pull_request:47306", - "right": "pull_request:47420", + "left": "pull_request:46604", + "right": "pull_request:47003", "accept": false, - "reason": "Neuron experimentation scripts and tokenizer special-token cleanup are unrelated." + "reason": "Mistral tekken utilities vs OpenVINO exporter; unrelated subsystems." }, { - "left": "pull_request:47360", - "right": "pull_request:47422", + "left": "pull_request:47082", + "right": "pull_request:47108", "accept": false, - "reason": "OTEL CI workflow changes and torchaudio pinning/RNNT fixes are unrelated." + "reason": "RoMa model support vs doc-builder pin comment cleanup; not the same issue." }, { - "left": "pull_request:47204", - "right": "pull_request:47422", + "left": "pull_request:46266", + "right": "pull_request:47216", "accept": false, - "reason": "Exporter-test Docker image changes are not the same as torchaudio pinning and RNN-T fixes." + "reason": "Dino modernization work vs compressed-tensors FP8 support; unrelated model/quantization work." }, { - "left": "pull_request:46912", - "right": "pull_request:47078", + "left": "pull_request:47078", + "right": "pull_request:47275", "accept": false, - "reason": "RoPE test updates and Japanese docs wording edits are unrelated." + "reason": "Japanese docs wording changes vs EfficientViT-SAM model support; different scopes." }, { - "left": "pull_request:47322", - "right": "pull_request:47420", - "accept": false, - "reason": "Quark config-class changes and tokenizer special-token cleanup are unrelated." - }, - { - "left": "pull_request:47179", - "right": "pull_request:47301", - "accept": false, - "reason": "Granite-SWA model support and MTP docs are unrelated." - }, - { - "left": "pull_request:47166", - "right": "pull_request:47422", + "left": "pull_request:47082", + "right": "pull_request:47336", "accept": false, - "reason": "Quantization test expectation updates and torchaudio pinning fixes are not the same issue." + "reason": "RoMa model support vs fully offloaded save bugfix; unrelated." }, { - "left": "pull_request:47166", - "right": "pull_request:47352", + "left": "pull_request:44517", + "right": "pull_request:45157", "accept": false, - "reason": "Quantization test updates and distributed runtime utilities are unrelated." + "reason": "Qwen3-TTS model support vs PrismML Bonsai support; separate model additions." }, { - "left": "pull_request:47298", - "right": "pull_request:47438", + "left": "pull_request:47210", + "right": "pull_request:47304", "accept": false, - "reason": "MI450 image-decoder/model failures and DSA mask materialization changes are different problems." + "reason": "Packed SDPA routing vs MPS graph-cache docs; unrelated." }, { "left": "pull_request:47275", - "right": "pull_request:47422", - "accept": false, - "reason": "EfficientViT-SAM support and CI torchaudio pinning are unrelated." - }, - { - "left": "pull_request:47298", - "right": "pull_request:47420", - "accept": false, - "reason": "MI450 image-decoder fixes and tokenizer special-token cleanup are unrelated." - }, - { - "left": "pull_request:47143", - "right": "pull_request:47320", - "accept": false, - "reason": "NeuCodec support and response_schema deprecation do not share the same code path." - }, - { - "left": "pull_request:46912", - "right": "pull_request:47438", - "accept": false, - "reason": "RoPE tests and less-materialized-mask changes are unrelated." - }, - { - "left": "pull_request:47105", - "right": "pull_request:47298", - "accept": false, - "reason": "Re-enabling VITS forward tests is unrelated to MI450 image decoding fixes." - }, - { - "left": "pull_request:47292", - "right": "pull_request:47383", - "accept": false, - "reason": "TipsV2 DPT XPU test expectation fixes and Kimi integration tests are unrelated." - }, - { - "left": "pull_request:45157", - "right": "pull_request:46658", - "accept": false, - "reason": "PrismML Bonsai support and Step-3.7 Flash model support are separate model additions." - }, - { - "left": "pull_request:47078", - "right": "pull_request:47322", + "right": "pull_request:47336", "accept": false, - "reason": "Japanese docs wording changes and Quark config changes are unrelated." + "reason": "EfficientViT-SAM model support vs offloaded model saving bugfix; different code paths." }, { - "left": "pull_request:44517", - "right": "pull_request:47301", + "left": "pull_request:46957", + "right": "pull_request:47216", "accept": false, - "reason": "Qwen3-TTS model support and MTP docs are unrelated." + "reason": "Finegrained FP8 kernel/runtime changes vs compressed-tensors FP8 support; both quantization-themed but not the same fix." }, { - "left": "pull_request:47294", - "right": "pull_request:47438", + "left": "pull_request:47216", + "right": "pull_request:47336", "accept": false, - "reason": "Kernel-native causal_conv1d utilities and less-materialized masks are different optimization efforts." + "reason": "Compressed-tensors FP8 support vs offloaded save bugfix; unrelated." }, { - "left": "pull_request:47113", - "right": "pull_request:47304", + "left": "pull_request:47306", + "right": "pull_request:47420", "accept": false, - "reason": "Audio backend requirements and MPS graph-cache docs are unrelated." + "reason": "Neuron experimentation scripts vs tokenizer special-token return cleanup; unrelated." }, { - "left": "pull_request:47294", - "right": "pull_request:47360", + "left": "pull_request:47204", + "right": "pull_request:47422", "accept": false, - "reason": "Kernel-native causal_conv1d functions and OTEL CI workflow changes are unrelated." + "reason": "Exporter-test Docker image setup vs CI torch/torchaudio pinning; both CI-related but different problems." } ] }, @@ -24988,7 +24987,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-40546-81", + "cluster_id": "cluster-40546-82", "nodes": [ "pull_request:41855", "pull_request:42668", @@ -24999,76 +24998,94 @@ "pull_request:43995", "pull_request:44517", "pull_request:44827", + "pull_request:45157", "pull_request:45170", - "pull_request:45213", "pull_request:45655", "pull_request:46048", + "pull_request:46196", "pull_request:46381", "pull_request:46604", "pull_request:46658", - "pull_request:46836", "pull_request:46895", "pull_request:46912", "pull_request:46957", - "pull_request:46958", - "pull_request:47003", "pull_request:47077", "pull_request:47078", - "pull_request:47090", "pull_request:47096", + "pull_request:47105", "pull_request:47113", "pull_request:47143", "pull_request:47163", "pull_request:47166", - "pull_request:47173", "pull_request:47175", + "pull_request:47179", "pull_request:47210", "pull_request:47243", "pull_request:47275", "pull_request:47291", - "pull_request:47292", "pull_request:47294", "pull_request:47298", "pull_request:47301", "pull_request:47304", "pull_request:47306", + "pull_request:47320", "pull_request:47322", - "pull_request:47352", "pull_request:47360", "pull_request:47367", "pull_request:47383", - "pull_request:47438" + "pull_request:47420", + "pull_request:47422", + "pull_request:47438", + "pull_request:47447" ], "soft_pairs": [ + "pull_request:47360|pull_request:47422", + "pull_request:46912|pull_request:47078", + "pull_request:47322|pull_request:47420", + "pull_request:47179|pull_request:47301", + "pull_request:47166|pull_request:47422", + "pull_request:47298|pull_request:47438", + "pull_request:47143|pull_request:47320", + "pull_request:47298|pull_request:47420", + "pull_request:46912|pull_request:47294", + "pull_request:47175|pull_request:47294", + "pull_request:47175|pull_request:47438", + "pull_request:46895|pull_request:47306", + "pull_request:46196|pull_request:47301", + "pull_request:47243|pull_request:47301", + "pull_request:47175|pull_request:47447", + "pull_request:46912|pull_request:47438", + "pull_request:47105|pull_request:47298", + "pull_request:47078|pull_request:47322", + "pull_request:45157|pull_request:46658", + "pull_request:44517|pull_request:47301", + "pull_request:47294|pull_request:47438", + "pull_request:47113|pull_request:47304", "pull_request:46912|pull_request:47291", - "pull_request:47163|pull_request:47352", + "pull_request:47438|pull_request:47447", "pull_request:43451|pull_request:44517", - "pull_request:47352|pull_request:47383", "pull_request:44517|pull_request:47383", "pull_request:44517|pull_request:47143", - "pull_request:43451|pull_request:47352", "pull_request:47078|pull_request:47243", + "pull_request:47210|pull_request:47291", "pull_request:46957|pull_request:47383", "pull_request:46604|pull_request:47163", - "pull_request:47210|pull_request:47291", "pull_request:46604|pull_request:47383", - "pull_request:47143|pull_request:47352", - "pull_request:46895|pull_request:47298", "pull_request:46957|pull_request:47360", "pull_request:47113|pull_request:47175", + "pull_request:46895|pull_request:47298", "pull_request:46381|pull_request:47322", "pull_request:46957|pull_request:47301", - "pull_request:43451|pull_request:46604", + "pull_request:47298|pull_request:47447", "pull_request:47113|pull_request:47275", "pull_request:46912|pull_request:47113", "pull_request:47077|pull_request:47367", - "pull_request:47360|pull_request:47438", "pull_request:47294|pull_request:47306", "pull_request:46658|pull_request:46895", - "pull_request:43451|pull_request:46381", - "pull_request:46381|pull_request:47096", "pull_request:47210|pull_request:47367", "pull_request:47275|pull_request:47367", + "pull_request:43451|pull_request:46381", + "pull_request:46381|pull_request:47096", "pull_request:41855|pull_request:43424", "pull_request:41855|pull_request:43823", "pull_request:42668|pull_request:43823", @@ -25083,328 +25100,342 @@ "pull_request:44827|pull_request:46048", "pull_request:43424|pull_request:43823", "pull_request:43251|pull_request:44827", - "pull_request:44827|pull_request:45655", - "pull_request:43995|pull_request:45213", - "pull_request:44827|pull_request:45213", - "pull_request:43995|pull_request:44827", - "pull_request:43823|pull_request:45213", - "pull_request:46895|pull_request:47090", - "pull_request:46958|pull_request:47090", - "pull_request:46895|pull_request:46958", - "pull_request:47173|pull_request:47292", - "pull_request:46912|pull_request:47175", - "pull_request:47096|pull_request:47383", - "pull_request:47301|pull_request:47304", - "pull_request:47077|pull_request:47304", - "pull_request:46658|pull_request:46836", - "pull_request:47210|pull_request:47301", - "pull_request:47163|pull_request:47166", - "pull_request:47003|pull_request:47243" + "pull_request:44827|pull_request:45655" ], - "prepared_review_unit_hash": "e8dd197482d486b96a7500d62947f24e3a094a128230bec6bcbd5f402baeaa2a", + "prepared_review_unit_hash": "6ebc71068d5fd3cc36c24c76d39682166ec2953e0530b31f366173ad49b91d32", "component_budget": { - "node_count": 81, - "item_count": 81, - "soft_pair_count": 225, - "serialized_chars": 279853, - "estimated_input_tokens": 69964, - "estimated_eval_tokens": 140184 + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 60, - "serialized_chars": 134699, - "estimated_input_tokens": 33675, - "estimated_eval_tokens": 67606 + "soft_pair_count": 62, + "serialized_chars": 131803, + "estimated_input_tokens": 32951, + "estimated_eval_tokens": 66158 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Only one real duplicate cluster stands out: the CohereASR training-loss double-shift fixes. The rest are soft-similarity false positives caused by broad model-addition, exporter, or test-only changes across unrelated code paths.", - "confidence": 0.94, + "summary": "This is a heterogeneous cluster of mostly unrelated PRs: model additions, docs edits, test tweaks, exporter/runtime fixes, CI changes, and small bug fixes. The soft-similarity pairs are topical at best and do not look like duplicate PRs for the same underlying change.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 47090 is the best canonical representative for the CohereASR double-shift cluster: it states the bug most explicitly and covers the same modeling/test path as 46895 and 46958.", + "canonical_pr_reason": "No single canonical PR: the items span unrelated subsystems and change types, so there is no duplicate-equivalent PR to anchor the cluster.", "best_issue_reason": null, - "best_pr_reason": "PR 47090 is the strongest overall PR candidate here because it is the clearest, most explicit fix in the only true duplicate set.", + "best_pr_reason": "If a representative PR had to be chosen, 47298 is the broadest infrastructure-oriented fix, but it still does not match the rest of the cluster as a duplicate.", "soft_edge_verdicts": [ + { + "left": "pull_request:47360", + "right": "pull_request:47422", + "accept": false, + "reason": "Different changes: OTEL workflow routing vs torchaudio pinning and RNN-T/CI fixes." + }, { "left": "pull_request:46912", - "right": "pull_request:47291", + "right": "pull_request:47078", "accept": false, - "reason": "Different changes: RoPE/test cleanup versus generation config deprecation; no shared bug." + "reason": "RoPE test coverage changes are unrelated to Japanese documentation wording edits." }, { - "left": "pull_request:47163", - "right": "pull_request:47352", + "left": "pull_request:47322", + "right": "pull_request:47420", "accept": false, - "reason": "Packed-attention XPU tests and distributed runtime utils are unrelated subsystems." + "reason": "Quark config-class rename is unrelated to a tokenizer special-token return cleanup." }, { - "left": "pull_request:43451", - "right": "pull_request:44517", + "left": "pull_request:47179", + "right": "pull_request:47301", "accept": false, - "reason": "Different model additions: Molmo2 versus Qwen3 TTS." + "reason": "Granite model support and MTP docs are different features, not the same fix." }, { - "left": "pull_request:47352", - "right": "pull_request:47383", + "left": "pull_request:47166", + "right": "pull_request:47422", "accept": false, - "reason": "Distributed runtime utils and Kimi integration tests are unrelated." + "reason": "Quantization test expectation updates are unrelated to torch/torchaudio pinning in CI." }, { - "left": "pull_request:44517", - "right": "pull_request:47383", + "left": "pull_request:47298", + "right": "pull_request:47438", "accept": false, - "reason": "Qwen3 TTS addition and Kimi integration tests are different changes." + "reason": "MI450 image/model failure fixes and less-materialized masks address different code paths." }, { - "left": "pull_request:44517", - "right": "pull_request:47143", + "left": "pull_request:47143", + "right": "pull_request:47320", "accept": false, - "reason": "Different model additions: Qwen3 TTS versus NeuCodec." + "reason": "NeuCodec model support is unrelated to deprecating the old response schema." }, { - "left": "pull_request:43451", - "right": "pull_request:47352", + "left": "pull_request:47298", + "right": "pull_request:47420", "accept": false, - "reason": "Molmo2 model addition and distributed runtime utils are unrelated." + "reason": "MI450 runtime fixes and tokenizer special-token cleanup are unrelated bugs." }, { - "left": "pull_request:47078", - "right": "pull_request:47243", + "left": "pull_request:46912", + "right": "pull_request:47294", "accept": false, - "reason": "Japanese docs wording and ExecuTorch exporter fixes do not match the same bug or change." + "reason": "RoPE test work and kernel-native causal_conv1d integration are unrelated changes." }, { - "left": "pull_request:46957", - "right": "pull_request:47383", + "left": "pull_request:47175", + "right": "pull_request:47294", "accept": false, - "reason": "Finegrained FP8 export/generation work and Kimi integration tests are unrelated." + "reason": "Olmo3 CI expectation updates do not overlap with kernel-native causal_conv1d support." }, { - "left": "pull_request:46604", - "right": "pull_request:47163", + "left": "pull_request:47175", + "right": "pull_request:47438", "accept": false, - "reason": "Mistral tekken tokenizer utilities and XPU paged-attention tests are unrelated." + "reason": "Test-value updates for Olmo3 are unrelated to the DSA mask optimization." }, { - "left": "pull_request:47210", - "right": "pull_request:47291", + "left": "pull_request:46895", + "right": "pull_request:47306", "accept": false, - "reason": "Packed SDPA varlen attention and CB config deprecation are different code paths." + "reason": "CohereASR double-shift fix and Neuron experiment scripts are different problems." }, { - "left": "pull_request:46604", - "right": "pull_request:47383", + "left": "pull_request:46196", + "right": "pull_request:47301", "accept": false, - "reason": "Mistral tekken tokenizer utilities and Kimi integration tests are unrelated." + "reason": "SDPA dispatch support for wav2vec2-family models is unrelated to MTP docs." }, { - "left": "pull_request:47143", - "right": "pull_request:47352", + "left": "pull_request:47243", + "right": "pull_request:47301", "accept": false, - "reason": "NeuCodec support and distributed runtime utils are unrelated." + "reason": "Executorch exporter fixes and assisted-decoding docs are not the same code-path issue." }, { - "left": "pull_request:46895", - "right": "pull_request:47298", + "left": "pull_request:47175", + "right": "pull_request:47447", "accept": false, - "reason": "CohereASR loss fix and MI450 image-loading/model failures are different bugs." + "reason": "Olmo3 test updates do not match speculative-decoding cache rollback changes." }, { - "left": "pull_request:46957", - "right": "pull_request:47360", + "left": "pull_request:46912", + "right": "pull_request:47438", "accept": false, - "reason": "Finegrained FP8 kernel/generation work and OTEL CI workflow changes are unrelated." + "reason": "RoPE test tuning and recurrent-attention mask reduction are unrelated." }, { - "left": "pull_request:47113", - "right": "pull_request:47175", + "left": "pull_request:47105", + "right": "pull_request:47298", "accept": false, - "reason": "Audio feature/proc requirement fixes and Olmo3 CI expectation updates are unrelated." + "reason": "Re-enabling a VITS test is unrelated to MI450 model failure fixes." }, { - "left": "pull_request:46381", + "left": "pull_request:47078", "right": "pull_request:47322", "accept": false, - "reason": "Tokenizers version/parser fixes and AMD Quark config changes are unrelated." + "reason": "Japanese docs wording changes are unrelated to Quark config API changes." }, { - "left": "pull_request:46957", + "left": "pull_request:45157", + "right": "pull_request:46658", + "accept": false, + "reason": "Prism GGUF support and Step-3.7 flash model work are different model/tooling efforts." + }, + { + "left": "pull_request:44517", "right": "pull_request:47301", "accept": false, - "reason": "Finegrained FP8 export/runtime work and MTP documentation are unrelated." + "reason": "Qwen3-TTS model support is unrelated to MTP documentation." }, { - "left": "pull_request:43451", - "right": "pull_request:46604", + "left": "pull_request:47294", + "right": "pull_request:47438", "accept": false, - "reason": "Molmo2 model addition and Mistral tokenizer utilities are unrelated." + "reason": "Kernel-native causal_conv1d functions and DSA mask changes fix different mechanisms." }, { "left": "pull_request:47113", - "right": "pull_request:47275", + "right": "pull_request:47320", "accept": false, - "reason": "Audio requirements fixes and EfficientViT-SAM model addition are unrelated." + "reason": "Audio feature/processor requirement fixes are unrelated to response-schema deprecation." }, { - "left": "pull_request:46912", - "right": "pull_request:47113", + "left": "pull_request:44517", + "right": "pull_request:47383", "accept": false, - "reason": "RoPE/model test updates and audio requirements fixes are different issues." + "reason": "Qwen3-TTS model addition and Kimi integration tests are unrelated." }, { - "left": "pull_request:47360", - "right": "pull_request:47438", + "left": "pull_request:44517", + "right": "pull_request:47143", "accept": false, - "reason": "OTEL CI workflow pilot and less-materialized-mask work are unrelated." + "reason": "Different model additions: Qwen3-TTS versus NeuCodec." }, { - "left": "pull_request:47294", - "right": "pull_request:47306", + "left": "pull_request:47078", + "right": "pull_request:47243", "accept": false, - "reason": "Kernel-native causal_conv1d work and Neuron training experiments are unrelated." + "reason": "Japanese docs wording edits are unrelated to Executorch exporter fixes." }, { - "left": "pull_request:46658", - "right": "pull_request:46895", + "left": "pull_request:47210", + "right": "pull_request:47291", "accept": false, - "reason": "Step3.7 model scaffolding and CohereASR loss-shift fix are unrelated." + "reason": "Packed SDPA routing and generation-config deprecation are unrelated features." }, { - "left": "pull_request:43451", - "right": "pull_request:46381", + "left": "pull_request:46957", + "right": "pull_request:47383", "accept": false, - "reason": "Molmo2 model addition and tokenizers dependency/parser fixes are unrelated." + "reason": "Finegrained FP8 decode-path tuning is unrelated to Kimi integration tests." }, { - "left": "pull_request:46381", - "right": "pull_request:47096", + "left": "pull_request:46604", + "right": "pull_request:47163", + "accept": false, + "reason": "Mistral tekken tokenizer utilities and XPU paged-attention tests are unrelated." + }, + { + "left": "pull_request:46604", + "right": "pull_request:47383", + "accept": false, + "reason": "Mistral tekken utilities and Kimi integration tests have no shared code-path." + }, + { + "left": "pull_request:46957", + "right": "pull_request:47360", + "accept": false, + "reason": "Finegrained FP8 kernel changes and OTEL CI workflow changes are unrelated." + }, + { + "left": "pull_request:47113", + "right": "pull_request:47275", + "accept": false, + "reason": "Audio requirement fixes do not overlap with EfficientVit-SAM model support." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47113", + "accept": false, + "reason": "RoPE testing changes and audio backend requirement fixes are unrelated." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47367", "accept": false, - "reason": "Tokenizers dependency/parser fixes and Kimi export work are unrelated." + "reason": "Japanese DistilBERT docs translation and pipeline device-placement clarification are unrelated docs updates." + }, + { + "left": "pull_request:47294", + "right": "pull_request:47306", + "accept": false, + "reason": "Kernel-native causal_conv1d functions do not match Neuron experiment scripts." }, { "left": "pull_request:47210", "right": "pull_request:47367", "accept": false, - "reason": "Packed SDPA attention routing and pipeline device placement are unrelated." + "reason": "Packed SDPA attention routing and pipeline device-placement docs are unrelated." }, { "left": "pull_request:47275", "right": "pull_request:47367", "accept": false, - "reason": "EfficientViT-SAM model addition and pipeline device placement are unrelated." + "reason": "EfficientVit-SAM model support is unrelated to pipeline device-placement clarification." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46381", + "accept": false, + "reason": "Molmo2 model addition and tokenizers version bump are unrelated." + }, + { + "left": "pull_request:46381", + "right": "pull_request:47096", + "accept": false, + "reason": "Tokenizer dependency/update work is unrelated to exportable Kimi support." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Mistral tokenizer missing methods and ExecuTorch export tests are unrelated." + "reason": "Mistral missing-methods work and Executorch dynamic-shape tests are different issues." }, { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Mistral tokenizer missing methods and MobileLLM addition are unrelated." + "reason": "Mistral tokenizer method additions are unrelated to MobileLLM model support." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "AudioFlamingo3 processor/auto-mapping changes and MobileLLM addition are unrelated." + "reason": "Processor-from-pretrained robustness and MobileLLM model addition are unrelated." }, { "left": "pull_request:43424", "right": "pull_request:47301", "accept": false, - "reason": "ExecuTorch export tests and MTP documentation are unrelated." + "reason": "Executorch dynamic-shape test coverage is unrelated to MTP docs." }, { - "left": "pull_request:43424", - "right": "pull_request:47383", + "left": "pull_request:43823", + "right": "pull_request:43995", "accept": false, - "reason": "ExecuTorch export tests and Kimi integration tests are unrelated." + "reason": "MobileLLM model support and Falcon output-interface refactor are different code paths." }, { - "left": "pull_request:43823", - "right": "pull_request:45655", + "left": "pull_request:43251", + "right": "pull_request:43823", "accept": false, - "reason": "MobileLLM addition and cls.config resolution ordering fix are unrelated." + "reason": "Cross-entropy kwargs passthrough fix is unrelated to MobileLLM support." }, { "left": "pull_request:44827", "right": "pull_request:47383", "accept": false, - "reason": "Mistral4 test fixes and Kimi integration tests are unrelated." + "reason": "Mistral4 test fixes and Kimi integration tests are unrelated model families." }, { "left": "pull_request:44827", - "right": "pull_request:45170", - "accept": false, - "reason": "Mistral4 tests and layrnorm->layernorm renaming target different bugs." - }, - { - "left": "pull_request:43424", - "right": "pull_request:44827", - "accept": false, - "reason": "ExecuTorch export tests and Mistral4 tests are unrelated." - }, - { - "left": "pull_request:43823", - "right": "pull_request:43995", + "right": "pull_request:45170", "accept": false, - "reason": "MobileLLM model addition and Falcon output-interface refactor are unrelated." + "reason": "Mistral4 test fixes and layrnorm-to-layernorm renaming are unrelated." }, { - "left": "pull_request:43251", - "right": "pull_request:43823", + "left": "pull_request:43424", + "right": "pull_request:44827", "accept": false, - "reason": "Cross-entropy kwarg forwarding and MobileLLM addition are unrelated." + "reason": "Executorch dynamic-shape export tests and Mistral4 test fixes are unrelated." }, { - "left": "pull_request:44827", + "left": "pull_request:43823", "right": "pull_request:46048", "accept": false, - "reason": "Mistral4 tests and TDT loss-kernel support are unrelated." + "reason": "MobileLLM model support and TDT loss kernel support are unrelated features." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "ExecuTorch export tests and MobileLLM addition are unrelated." + "reason": "Executorch export tests and MobileLLM model support do not address the same bug." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Cross-entropy kwarg forwarding and Mistral4 test fixes are unrelated." + "reason": "Cross-entropy kwargs handling and Mistral4 test updates are unrelated." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Mistral4 tests and cls.config resolution ordering are unrelated." - }, - { - "left": "pull_request:43995", - "right": "pull_request:45213", - "accept": false, - "reason": "Falcon output-interface refactor and model-creation skill scaffolding are unrelated." - }, - { - "left": "pull_request:44827", - "right": "pull_request:45213", - "accept": false, - "reason": "Mistral4 tests and model-creation skill scaffolding are unrelated." - }, - { - "left": "pull_request:43823", - "right": "pull_request:45213", - "accept": false, - "reason": "MobileLLM addition and model-creation skill scaffolding are unrelated." + "reason": "Mistral4 test fixes and config-resolution-order changes are unrelated issues." } ] }, @@ -25426,21 +25457,30 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-40546-81", + "cluster_id": "cluster-40546-82", "nodes": [ "pull_request:40546", "pull_request:43451", + "pull_request:43823", + "pull_request:43995", + "pull_request:44827", "pull_request:45157", + "pull_request:45213", "pull_request:46196", "pull_request:46509", "pull_request:46658", "pull_request:46836", + "pull_request:46895", "pull_request:46912", + "pull_request:46958", "pull_request:47003", + "pull_request:47077", + "pull_request:47090", "pull_request:47096", "pull_request:47143", "pull_request:47163", "pull_request:47166", + "pull_request:47173", "pull_request:47175", "pull_request:47210", "pull_request:47216", @@ -25459,30 +25499,44 @@ "pull_request:47387" ], "soft_pairs": [ + "pull_request:43995|pull_request:45213", + "pull_request:44827|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:43823|pull_request:45213", + "pull_request:46895|pull_request:47090", + "pull_request:46958|pull_request:47090", + "pull_request:46895|pull_request:46958", + "pull_request:47173|pull_request:47292", + "pull_request:46912|pull_request:47175", + "pull_request:47096|pull_request:47383", + "pull_request:47301|pull_request:47304", + "pull_request:47077|pull_request:47304", + "pull_request:46658|pull_request:46836", + "pull_request:47210|pull_request:47301", + "pull_request:47163|pull_request:47166", + "pull_request:47003|pull_request:47243", "pull_request:46509|pull_request:47143", "pull_request:46509|pull_request:47163", "pull_request:45157|pull_request:46836", - "pull_request:46836|pull_request:47387", "pull_request:47163|pull_request:47292", + "pull_request:46836|pull_request:47387", "pull_request:47291|pull_request:47322", - "pull_request:47301|pull_request:47336", "pull_request:45157|pull_request:46509", - "pull_request:47301|pull_request:47383", + "pull_request:47301|pull_request:47336", "pull_request:47216|pull_request:47383", "pull_request:47336|pull_request:47383", "pull_request:46509|pull_request:47387", "pull_request:47304|pull_request:47367", + "pull_request:47301|pull_request:47383", "pull_request:47216|pull_request:47301", "pull_request:46912|pull_request:47298", "pull_request:46509|pull_request:47301", - "pull_request:43451|pull_request:46836", "pull_request:47175|pull_request:47298", "pull_request:47163|pull_request:47210", - "pull_request:47294|pull_request:47298", "pull_request:45157|pull_request:47387", + "pull_request:47294|pull_request:47298", "pull_request:46658|pull_request:47387", "pull_request:46196|pull_request:47143", - "pull_request:47143|pull_request:47243", "pull_request:47143|pull_request:47163", "pull_request:46658|pull_request:47143", "pull_request:46658|pull_request:47322", @@ -25503,7 +25557,6 @@ "pull_request:47003|pull_request:47096", "pull_request:43451|pull_request:47322", "pull_request:47096|pull_request:47322", - "pull_request:43451|pull_request:47298", "pull_request:43451|pull_request:47306", "pull_request:47096|pull_request:47306", "pull_request:47003|pull_request:47306", @@ -25511,340 +25564,382 @@ "pull_request:47003|pull_request:47387", "pull_request:47096|pull_request:47143" ], - "prepared_review_unit_hash": "4f0ce8baeba436a30211923286ef49c4970d46b0334a56f8ad61403b9052fdc4", + "prepared_review_unit_hash": "cd92e37966c4c162686472aef59ca4a962a7ab140c18de0456f16e386debf2a5", "component_budget": { - "node_count": 81, - "item_count": 81, - "soft_pair_count": 225, - "serialized_chars": 279853, - "estimated_input_tokens": 69964, - "estimated_eval_tokens": 140184 + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 }, "budget": { - "node_count": 29, - "item_count": 29, - "soft_pair_count": 51, - "serialized_chars": 85123, - "estimated_input_tokens": 21281, - "estimated_eval_tokens": 42818 + "node_count": 38, + "item_count": 38, + "soft_pair_count": 64, + "serialized_chars": 112164, + "estimated_input_tokens": 28041, + "estimated_eval_tokens": 56338 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Heterogeneous cluster: mostly unrelated model additions, exporter changes, test-only fixes, and docs updates. No soft-edge pair looks like the same underlying bug or change, so all candidate duplicates are rejected.", - "confidence": 0.93, + "summary": "This cluster is mostly a soft-similarity grab bag of unrelated PRs that share broad subsystems or test-style changes. The only true duplicate family is the CohereASR training-loss double-shift fix trio; everything else should be kept separate. I would use PR #47090 as the representative canonical PR for that duplicate family.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43451 is the most complete self-contained feature addition in the set (docs, model code, tests, and auto-mapping updates), so it is the best representative PR if one must be chosen; however, this cluster is not actually a duplicate set.", + "canonical_pr_reason": "PR #47090 is the clearest final form of the only real duplicate set here: it targets the same CohereASR double-shift bug as #46895 and #46958, touches the same files, and states the intended label alignment most explicitly.", "best_issue_reason": null, - "best_pr_reason": "PR 43451 is the broadest and most complete standalone change among the items, with the clearest end-to-end implementation and validation. It is the best representative PR, but not a duplicate canonical in this heterogeneous group.", + "best_pr_reason": "PR #47090 is the best overall representative because it is the most explicit and polished version of the same concrete fix, while the other strong-similarity pairs in the cluster are unrelated false positives.", "soft_edge_verdicts": [ { - "left": "pull_request:46509", - "right": "pull_request:47143", + "left": "pull_request:43995", + "right": "pull_request:45213", "accept": false, - "reason": "Both add new model support, but for different models (Nemotron Omni vs NeuCodec) with distinct codepaths and files." + "reason": "Falcon output-collection refactor vs a model-creation skill PR; unrelated changes." }, { - "left": "pull_request:46509", - "right": "pull_request:47163", + "left": "pull_request:44827", + "right": "pull_request:45213", "accept": false, - "reason": "Nemotron Omni model support is unrelated to XPU paged-attention test coverage." + "reason": "Mistral4 test/export fix vs model-creation skill; different model and codepath." }, { - "left": "pull_request:45157", + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Different models and different fixes; only broad transformer-internals similarity." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "MobileLLM scaffold vs model-creation skill; not the same underlying change." + }, + { + "left": "pull_request:46895", + "right": "pull_request:47090", + "accept": true, + "reason": "Same CohereASR double-shift bug, same files, same issue target; 47090 is the later, clearer variant." + }, + { + "left": "pull_request:46958", + "right": "pull_request:47090", + "accept": true, + "reason": "Same CohereASR training-loss bug and codepath; 47090 is just the more explicit final fix." + }, + { + "left": "pull_request:46895", + "right": "pull_request:46958", + "accept": true, + "reason": "Both fix the same CohereASR loss double-shift bug in the same files." + }, + { + "left": "pull_request:47173", + "right": "pull_request:47292", + "accept": false, + "reason": "Both are XPU test expectation updates, but for different models/files and different failures." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47175", + "accept": false, + "reason": "RoPE-related test updates vs Olmo3 CI expectation tweaks; not the same bug." + }, + { + "left": "pull_request:47096", + "right": "pull_request:47383", + "accept": false, + "reason": "Kimi exporter/model work vs Kimi integration tests; related project area but different concrete changes." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47304", + "accept": false, + "reason": "MTP docs vs MPS graph-cache docs; unrelated documentation changes." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47304", + "accept": false, + "reason": "Japanese DistilBERT docs vs MPS graph-cache docs; unrelated." + }, + { + "left": "pull_request:46658", "right": "pull_request:46836", "accept": false, - "reason": "PrismML Bonsai GGUF support and Unlimited OCR are unrelated changes in different subsystems." + "reason": "Step3.7 Flash model conversion work vs Unlimited OCR model support; different model additions." }, { - "left": "pull_request:46836", - "right": "pull_request:47387", + "left": "pull_request:47210", + "right": "pull_request:47301", "accept": false, - "reason": "Unlimited OCR model support is unrelated to one-to-many conversion support in core model loading." + "reason": "Packed-SDPA runtime change vs MTP docs; not the same change." }, { "left": "pull_request:47163", - "right": "pull_request:47292", + "right": "pull_request:47166", + "accept": false, + "reason": "Paged-attention XPU tests vs GPTQ/MXFP4 test tweaks; different test fixes." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47243", "accept": false, - "reason": "Both are XPU-related test changes, but they target different models and different failure modes." + "reason": "OpenVINO exporter addition vs Executorch exporter fixes; different exporter backends." }, { "left": "pull_request:47291", "right": "pull_request:47322", "accept": false, - "reason": "Different quantization/config refactors: deprecated continuous batching config vs AMD Quark config class rename." + "reason": "Generation-config deprecation vs AMD Quark config rename; unrelated quantization/config changes." }, { - "left": "pull_request:47301", - "right": "pull_request:47336", + "left": "pull_request:45157", + "right": "pull_request:46836", + "accept": false, + "reason": "Prism GGUF dequant support vs Unlimited OCR model support; unrelated model/format work." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47292", + "accept": false, + "reason": "Paged-attention XPU tests vs tipsv2_dpt XPU tests; distinct model failures." + }, + { + "left": "pull_request:46836", + "right": "pull_request:47387", "accept": false, - "reason": "Docs for MTP support and a model-saving offload bugfix are unrelated changes." + "reason": "Unlimited OCR model support vs one-to-many conversion infrastructure; not the same change." }, { "left": "pull_request:45157", "right": "pull_request:46509", "accept": false, - "reason": "Different new model additions with no shared bug or codepath." + "reason": "Prism GGUF support vs Nemotron Omni model support; unrelated." }, { "left": "pull_request:47301", - "right": "pull_request:47383", + "right": "pull_request:47336", "accept": false, - "reason": "MTP documentation and Kimi integration tests are unrelated." + "reason": "MTP docs vs fully offloaded model saving bugfix; unrelated." }, { "left": "pull_request:47216", "right": "pull_request:47383", "accept": false, - "reason": "Compressed-tensors FP8 support is unrelated to Kimi integration tests." + "reason": "Compressed-tensors fp8 work vs Kimi integration tests; different changes." }, { - "left": "pull_request:47336", - "right": "pull_request:47383", + "left": "pull_request:47163", + "right": "pull_request:47210", "accept": false, - "reason": "Model offloading save fix and Kimi integration tests are unrelated." + "reason": "Paged-attention XPU tests vs packed-SDPA runtime change; unrelated." }, { - "left": "pull_request:46509", - "right": "pull_request:47387", + "left": "pull_request:47003", + "right": "pull_request:47322", "accept": false, - "reason": "Nemotron Omni model support and one-to-many conversion support are unrelated." + "reason": "OpenVINO exporter vs Quark config rename; unrelated." }, { - "left": "pull_request:47304", - "right": "pull_request:47367", + "left": "pull_request:47096", + "right": "pull_request:47387", "accept": false, - "reason": "MPS graph-cache docs cleanup is unrelated to pipeline device-placement clarification." + "reason": "Kimi exporter/model work vs one-to-many conversion support; different codepaths." }, { - "left": "pull_request:47216", - "right": "pull_request:47301", + "left": "pull_request:47304", + "right": "pull_request:47367", "accept": false, - "reason": "Compressed-tensors FP8 work and MTP docs are unrelated." + "reason": "MPS graph-cache docs vs pipeline device-placement clarification; unrelated docs." }, { - "left": "pull_request:46912", - "right": "pull_request:47298", + "left": "pull_request:47301", + "right": "pull_request:47383", "accept": false, - "reason": "Both touch testing, but RoPE test updates and MI450 model-failure fixes address different issues." + "reason": "MTP docs vs Kimi integration tests; unrelated." }, { - "left": "pull_request:46509", + "left": "pull_request:47216", "right": "pull_request:47301", "accept": false, - "reason": "Different domains: model support vs generation docs." - }, - { - "left": "pull_request:43451", - "right": "pull_request:46836", - "accept": false, - "reason": "Molmo2 model addition is unrelated to Unlimited OCR." + "reason": "Compressed-tensors fp8 work vs MTP docs; unrelated." }, { - "left": "pull_request:47175", + "left": "pull_request:46912", "right": "pull_request:47298", "accept": false, - "reason": "Olmo3 CI expectation updates and MI450 failures are different model/test fixes." + "reason": "RoPE test tuning vs MI450 image-decoding/model failures; different subsystems." }, { "left": "pull_request:47163", "right": "pull_request:47210", "accept": false, - "reason": "Paged-attention XPU tests and SDPA varlen packed attention are different changes." + "reason": "Paged-attention test updates vs packed-SDPA runtime path; not the same change." }, { - "left": "pull_request:47294", - "right": "pull_request:47298", + "left": "pull_request:45157", + "right": "pull_request:47387", "accept": false, - "reason": "Kernel-function plumbing for causal_conv1d is unrelated to MI450 image-loading/model failures." + "reason": "Prism GGUF support vs one-to-many conversion support; unrelated." }, { - "left": "pull_request:45157", - "right": "pull_request:47387", + "left": "pull_request:47294", + "right": "pull_request:47298", "accept": false, - "reason": "PrismML GGUF quantization support and one-to-many conversion support are unrelated." + "reason": "Kernel-native causal_conv1d support vs MI450 image-decoder fixes; different codepaths." }, { "left": "pull_request:46658", "right": "pull_request:47387", "accept": false, - "reason": "Step-3.7 Flash model work is unrelated to core model-loading conversion changes." + "reason": "Step3.7 Flash model work vs one-to-many conversion infrastructure; unrelated." }, { "left": "pull_request:46196", "right": "pull_request:47143", "accept": false, - "reason": "wav2vec2 SDPA dispatch support and NeuCodec model support do not fix the same problem." - }, - { - "left": "pull_request:47143", - "right": "pull_request:47243", - "accept": false, - "reason": "NeuCodec model support and executorch exporter fixes are different codepaths and goals." + "reason": "SDPA dispatch support vs NeuCodec model support; different model areas." }, { "left": "pull_request:47143", "right": "pull_request:47163", "accept": false, - "reason": "NeuCodec model support is unrelated to XPU paged-attention testing." + "reason": "NeuCodec model support vs paged-attention XPU tests; unrelated." }, { "left": "pull_request:46658", "right": "pull_request:47143", "accept": false, - "reason": "Step-3.7 Flash implementation and NeuCodec support are unrelated model additions." + "reason": "Step3.7 Flash model work vs NeuCodec model support; unrelated." }, { "left": "pull_request:46658", "right": "pull_request:47322", "accept": false, - "reason": "Step-3.7 Flash support and AMD Quark config changes are unrelated." + "reason": "Step3.7 model work vs Quark config changes; unrelated." }, { "left": "pull_request:47143", "right": "pull_request:47387", "accept": false, - "reason": "NeuCodec model support is unrelated to one-to-many conversion support." + "reason": "NeuCodec model support vs one-to-many conversion support; unrelated." }, { "left": "pull_request:47143", "right": "pull_request:47322", "accept": false, - "reason": "NeuCodec model support and Quark config changes are unrelated." + "reason": "NeuCodec model support vs Quark config rename; unrelated." }, { "left": "pull_request:45157", "right": "pull_request:47143", "accept": false, - "reason": "Different model-support PRs for different architectures; no shared bug or mergeable common fix." + "reason": "Prism GGUF support vs NeuCodec model support; unrelated." }, { "left": "pull_request:47301", "right": "pull_request:47367", "accept": false, - "reason": "Generation docs for MTP and pipeline device docs are unrelated." + "reason": "MTP docs vs pipeline device-placement docs; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47291", "accept": false, - "reason": "Molmo2 model support and deprecating continuous batching config are unrelated." + "reason": "Molmo2 model addition vs generation-config deprecation; unrelated." }, { "left": "pull_request:47166", "right": "pull_request:47383", "accept": false, - "reason": "Quantization test updates and Kimi integration tests are unrelated." + "reason": "GPTQ/MXFP4 test tweaks vs Kimi integration tests; different failures." }, { "left": "pull_request:40546", "right": "pull_request:43451", "accept": false, - "reason": "VibeVoice and Molmo2 are distinct model additions with no shared change." - }, - { - "left": "pull_request:47306", - "right": "pull_request:47367", - "accept": false, - "reason": "Neuron experiments and pipeline device docs are unrelated." + "reason": "VibeVoice and Molmo2 are separate model-addition PRs." }, { "left": "pull_request:43451", "right": "pull_request:47143", "accept": false, - "reason": "Molmo2 and NeuCodec are separate model-support PRs for different models." + "reason": "Molmo2 vs NeuCodec; different models and artifacts." }, { "left": "pull_request:43451", "right": "pull_request:47163", "accept": false, - "reason": "Molmo2 model support is unrelated to XPU paged-attention tests." + "reason": "Molmo2 model addition vs paged-attention XPU tests; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47383", "accept": false, - "reason": "Molmo2 model addition and Kimi integration tests are unrelated." + "reason": "Molmo2 vs Kimi integration tests; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:46509", "accept": false, - "reason": "Molmo2 and Nemotron Omni are different model additions." + "reason": "Molmo2 vs Nemotron Omni; separate model additions." }, { "left": "pull_request:43451", "right": "pull_request:47096", "accept": false, - "reason": "Molmo2 model support is unrelated to Kimi exportability work." + "reason": "Molmo2 vs Kimi exporter/model work; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47003", "accept": false, - "reason": "Molmo2 model support and OpenVINO exporter work are unrelated." + "reason": "Molmo2 vs OpenVINO exporter work; unrelated." }, { "left": "pull_request:47003", "right": "pull_request:47096", "accept": false, - "reason": "OpenVINO exporter infrastructure and exportable Kimi model support are different changes." + "reason": "OpenVINO exporter addition vs Kimi exporter/model work; different exporter paths." }, { - "left": "pull_request:43451", - "right": "pull_request:47322", + "left": "pull_request:47003", + "right": "pull_request:47306", "accept": false, - "reason": "Molmo2 support and Quark config changes are unrelated." + "reason": "OpenVINO exporter vs Neuron experiments; unrelated." }, { "left": "pull_request:47096", "right": "pull_request:47322", "accept": false, - "reason": "Exportability fixes for Kimi and AMD Quark config changes do not address the same issue." - }, - { - "left": "pull_request:43451", - "right": "pull_request:47298", - "accept": false, - "reason": "Molmo2 model addition is unrelated to MI450 image-utils/model failures." + "reason": "Kimi exporter/model work vs Quark config rename; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47306", "accept": false, - "reason": "Molmo2 model support and Neuron experimentation are unrelated." - }, - { - "left": "pull_request:47096", - "right": "pull_request:47306", - "accept": false, - "reason": "Kimi exporter support and Neuron experiments are unrelated." - }, - { - "left": "pull_request:47003", - "right": "pull_request:47306", - "accept": false, - "reason": "OpenVINO exporter work and Neuron experiments are unrelated." + "reason": "Molmo2 model addition vs Neuron experiments; unrelated." }, { "left": "pull_request:47096", "right": "pull_request:47387", "accept": false, - "reason": "Kimi exportability work and one-to-many conversion support are unrelated." + "reason": "Kimi exporter/model work vs one-to-many conversion support; unrelated." }, { "left": "pull_request:47003", "right": "pull_request:47387", "accept": false, - "reason": "OpenVINO exporter work and one-to-many conversion support are unrelated." + "reason": "OpenVINO exporter vs one-to-many conversion support; unrelated." }, { "left": "pull_request:47096", "right": "pull_request:47143", "accept": false, - "reason": "Kimi exportability work and NeuCodec support are unrelated." + "reason": "Kimi exporter/model work vs NeuCodec model support; unrelated." } ] }, @@ -25874,7 +25969,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "c52bd3bb28b8e4f35be578ee254929a86047782a6b9fc9d970415af57691ec6d", + "prepared_review_unit_hash": "051aca41be6b9d5e295f3b976691a72e2b5372a742e36c36eea5db459af73c95", "component_budget": { "node_count": 2, "item_count": 2, @@ -25896,18 +25991,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are unrelated: one adds covariate forecasting support to TimesFM, the other adds DINOv3 image classification support. They do not describe the same bug or change and should not be clustered as duplicates.", + "summary": "These are two unrelated feature PRs: one adds TimesFM covariate forecasting, the other adds DINOv3 image classification support. They do not appear to fix the same code-path problem or represent the same change, so they should not be clustered as duplicates.", "confidence": 0.99, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR #41224 is the more self-contained change: it adds a specific DINOv3 image-classification capability with matching model, auto-model, docs, and tests. It is a cleaner representative of a single coherent feature.", "best_issue_reason": null, - "best_pr_reason": "No single PR is a good canonical representative because the cluster contains two unrelated changes in different model families and code paths.", + "best_pr_reason": "PR #41224 is the strongest standalone PR in the set because it has a narrower, clearly defined scope and a concrete end-user feature implementation. PR #40755 is also valid, but it is a different feature area and not a better cluster representative.", "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different models and different functionality: TimesFM covariate forecasting vs DINOv3 image classification. No shared code path or mergeable combined change." + "reason": "Different model families and different features: TimesFM covariate forecasting vs DINOv3 image classification support. No shared bug or mergeable code-path overlap." } ] }, @@ -26000,12 +26095,12 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "2614590065f61c97ec496d0eb818183edd71094b518a1cb0457139fbd1b9b4d1", + "prepared_review_unit_hash": "1a480bee2bdeb61df7a9ac95e70cf165714f147be5a33273d2cc2a566aa7d204", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5934, + "serialized_chars": 5935, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -26013,7 +26108,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5934, + "serialized_chars": 5935, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -26022,18 +26117,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two GitHub Actions upgrade PRs overlap in workflows but target different action families and motivations, so they are not duplicates and should not be merged.", - "confidence": 0.91, + "summary": "Both pull requests are broad GitHub Actions dependency refreshes, but they target different concrete action sets: PR 42975 is specifically about Node 24 compatibility for core workflow actions, while PR 42976 upgrades docker/build-push-action and a few other actions to latest versions. They overlap in affected workflows, but not in the same underlying change, so this is not a duplicate pair.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 42975 is the better canonical PR because it has the clearer, narrower purpose and directly targets the Node 24 compatibility work across workflow actions.", "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the better umbrella candidate because it explicitly targets Node 24 compatibility across many workflows, whereas 42976 is a narrower docker/build-push-action version bump.", + "best_pr_reason": "PR 42975 is the strongest representative of the cluster: the title matches the main maintenance goal, and the diff shows a coherent set of action-version bumps for compatibility rather than a mixed latest-version refresh.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Both update workflow dependencies, but they change different actions (checkout/cache/upload-artifact vs docker/build-push-action) and solve different upgrade goals, so they are not the same underlying PR." + "reason": "Related workflow-maintenance changes, but they update different actions for different purposes. They do not appear to fix the same concrete code-path problem and would not naturally merge into one PR." } ] }, @@ -26068,7 +26163,7 @@ "pull_request:43448|pull_request:45490", "pull_request:45490|pull_request:46562" ], - "prepared_review_unit_hash": "27e039afffcc2ada84feaa9cc4abe4e763671a79be6ec8b022dac3c373f7a53a", + "prepared_review_unit_hash": "9ed2f37d53bbabe4824b118702b97dda15d11aa072275159a623912728b148da", "component_budget": { "node_count": 4, "item_count": 4, @@ -26090,36 +26185,36 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are four superficially similar but substantively different model-addition PRs (ViT NEPA, Molmo, CTSM, Step3p7). They share framework plumbing like docs and auto-registry files, but each targets a different model/codepath, so none should be clustered as duplicates.", - "confidence": 0.97, + "summary": "These are four separate model-addition pull requests (ViT NEPA, Molmo, CTSM, Step3p7). They overlap mainly in common integration files like docs and auto-mapping, but they implement different models and unrelated code paths.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: each item adds a different model and associated integration, not the same underlying change.", + "canonical_pr_reason": "PR 45490 is the best representative of the cluster because it is a substantial standalone model addition and has the strongest shared integration footprint across the set, but it is still not the same change as the others.", "best_issue_reason": null, - "best_pr_reason": "If a representative PR is needed, 46562 is the broadest standalone model-addition change, but it is still unrelated to the others as a duplicate target.", + "best_pr_reason": "PR 45490 is the best single PR to keep as the cluster representative: it is fully self-contained, touches core integration files, and has the highest discussion activity among the candidates.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different model additions (ViT NEPA vs Molmo); shared auto/docs files are just integration boilerplate." + "reason": "Different model additions (ViT NEPA vs Molmo). Shared docs/auto-registry files are expected for new model onboarding and do not indicate the same underlying change." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "Different architectures and codepaths; only generic registration/docs overlap." + "reason": "These add different models and modify different model implementations; overlap is only in shared registry/docs plumbing." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo and CTSM are separate model integrations, not the same bug or change." + "reason": "Molmo and CTSM are separate model integrations with distinct modeling, processing, and conversion code, so they are not mergeable as one PR." }, { "left": "pull_request:45490", "right": "pull_request:46562", "accept": false, - "reason": "CTSM vs Step3p7 are distinct model additions with different implementation surfaces." + "reason": "CTSM and Step3p7 are unrelated model additions; shared auto-mapping and docs edits are boilerplate for new model support, not evidence of the same bug or feature." } ] }, @@ -26168,46 +26263,130 @@ "pull_request:46539|pull_request:46922", "pull_request:46886|pull_request:46922" ], - "prepared_review_unit_hash": "6dd3099fde00aa17b2cebd1ff8b42f4802d2bb197ae31cf46261cfe79edd2c2c", + "prepared_review_unit_hash": "df7bce939e7bef926268497af0c6590c5265ed7bc24199864cbc80536409f5f1", "component_budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 15, - "serialized_chars": 19595, - "estimated_input_tokens": 4899, - "estimated_eval_tokens": 10054 + "serialized_chars": 19604, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 }, "budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 15, - "serialized_chars": 19595, - "estimated_input_tokens": 4899, - "estimated_eval_tokens": 10054 + "serialized_chars": 19604, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly unrelated small PRs, but the YOLOS pair (46886, 46922) targets the same device-mismatch bug under `device_map=\"auto\"` and overlaps on the same issue. The rest touch different models or unrelated test/CI areas and should stay separate.", - "confidence": 0.86, + "summary": "This cluster is mostly unrelated PRs spanning Whisper tests, CLIP/ViT tests, FP8 dequantization, generation watermarking, Phimoe CI, and YOLOS. The only true duplicate-like pair is the two YOLOS fixes, which both address the same device_map=\"auto\" device mismatch path in the same file.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46922 is the more complete YOLOS fix: it addresses the concrete runtime mismatch in the model forward path and also preserves the split-prevention change. It is the better representative of the underlying bug than the narrower 46886 patch.", + "canonical_pr_reason": "pull_request:46922 is the strongest representative because it includes the concrete YOLOS runtime fix plus the device-splitting safeguard, directly tied to the explicit issue target.", "best_issue_reason": null, - "best_pr_reason": "PR 46922 is the strongest standalone fix in the cluster because it directly patches the failing code path (`hidden_states + interpolated_mid_position_embeddings[i]`) and includes the device-map safeguard as part of the same change.", + "best_pr_reason": "pull_request:46922 is the best overall PR: it has the clearest concrete bug fix, a direct issue target, and the broadest evidence of addressing the underlying YOLOS device mismatch.", "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Different models and different kinds of changes; both are test-oriented but do not fix the same underlying bug or change." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper tokenizer test cleanup is unrelated to the FP8 dequant/load-path fix." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46380", + "accept": false, + "reason": "Whisper test changes and generation/watermarking validation changes are unrelated." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46922", + "accept": false, + "reason": "Whisper tests and YOLOS device mismatch fixes are unrelated code paths." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46539", + "accept": false, + "reason": "Test edits in CLIP/ViT are unrelated to the Phimoe CI/integration test fix." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Different subsystems and different fixes; no shared underlying bug." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46380", + "accept": false, + "reason": "CLIP/ViT test changes are unrelated to generation config validation." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46922", + "accept": false, + "reason": "Different models and issues; no common bug or mergeable unified fix." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46539", + "accept": false, + "reason": "FP8 dequantization changes and Phimoe CI environment changes are unrelated." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46380", + "accept": false, + "reason": "FP8 quantization/dequant logic is unrelated to generation config watermarking validation." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46922", + "accept": false, + "reason": "No shared code path or underlying bug; FP8 fix and YOLOS device mismatch are unrelated." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46539", + "accept": false, + "reason": "Generation config validation and CI/device memory patching solve different problems." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46922", + "accept": false, + "reason": "Unrelated generation config/watermarking fix vs YOLOS runtime device mismatch." + }, + { + "left": "pull_request:46539", + "right": "pull_request:46922", + "accept": false, + "reason": "Phimoe CI test infrastructure changes are unrelated to YOLOS model device placement." + }, { "left": "pull_request:46886", "right": "pull_request:46922", "accept": true, - "reason": "Both PRs address the same YOLOS device-mismatch bug under `device_map=\"auto\"` in `src/transformers/models/yolos/modeling_yolos.py`. 46886 only adds `_no_split_modules`, while 46922 fixes the forward-path tensor device placement and also includes the same split-prevention change, so they are mergeable as one underlying fix." + "reason": "Both PRs fix the same YOLOS device_map=\"auto\" device mismatch in the same model file, and the changes are plausibly mergeable into one PR." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays appropriately conservative. The YOLOS pair is supported by shared issue target 46869 and the diffs show 46886 adds the split-prevention list while 46922 adds the device-placement fix plus the same list, so treating them as mergeable pieces of one bugfix is reasonable. The rest of the cluster is correctly characterized as unrelated model/test/CI work." + "feedback": "The summary is grounded in the packet, and the negative pair verdicts stay conservative. The only accepted soft-edge pair is the YOLOS pair, which is reasonably mergeable into one PR for the same bugfix given the shared file and issue target." }, "evaluator_used": true, "retried": false, @@ -26237,12 +26416,12 @@ "pull_request:43488|pull_request:43532", "pull_request:43532|pull_request:46747" ], - "prepared_review_unit_hash": "2c46c3cf9534910b251107c3063873656a7ba6d12f87b15a1c8eebe7d38258ba", + "prepared_review_unit_hash": "d32c47f3cef7a6a16a99a100902f894dd5c04c28172130460a21e671dfbf0528", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 6348, + "serialized_chars": 6347, "estimated_input_tokens": 1587, "estimated_eval_tokens": 3430 }, @@ -26250,7 +26429,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 6348, + "serialized_chars": 6347, "estimated_input_tokens": 1587, "estimated_eval_tokens": 3430 }, @@ -26259,30 +26438,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three PRs are all unrelated '[don't merge]' demo/test changes that only share superficial CI/test overlap. They do not fix the same concrete bug or implement the same change, so none should be clustered as duplicates.", - "confidence": 0.96, + "summary": "The cluster contains three unrelated draft/don't-merge PRs. They share some weak topical overlap around repo automation and tests, but they do not fix the same concrete bug or implement the same change.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No single canonical PR: 43488 is a packaging/dependency/version-bump style change, 43532 is a workflow/tooling demo touching several CI files, and 46747 is a distinct security-gate workflow trigger change.", "best_issue_reason": null, - "best_pr_reason": "PR 43532 is the broadest representative of the set because it touches the most files and the largest CI/testing surface, but it is still not a duplicate of the other PRs.", + "best_pr_reason": "PR 46747 is the most self-contained and narrowly scoped change, so it is the best representative of the cluster, even though it is not a duplicate of the others.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:46747", "accept": false, - "reason": "One changes packaging/versioning and a test file; the other changes a security-gate workflow. Different code paths and different purposes." + "reason": "Different changes: 43488 touches setup/dependency metadata and a test, while 46747 changes a CI workflow trigger/permissions path. No shared concrete bug or fix." }, { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "Only superficial overlap in test/CI areas. 43488 is about setup.py/dependency/version changes, while 43532 rewires comment-driven CI workflows and testing utilities." + "reason": "43488 is about packaging/deps/versioning; 43532 edits workflow files plus testing/notification utilities. The shared test file is incidental and not the same underlying change." }, { "left": "pull_request:43532", "right": "pull_request:46747", "accept": false, - "reason": "Both touch GitHub workflows, but they modify different workflows for different behavior. Not the same underlying bug or change, and not something that would plausibly be merged into one PR." + "reason": "Both involve workflows, but they target different CI systems and problems: 43532 is a diff-display/comment-CI demo, while 46747 changes a security-gate workflow trigger. Not mergeable as one PR." } ] }, @@ -26312,7 +26491,7 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "526b1fe3f32f94327add73f8ce7ab8b239b6e72840b894ddcf9fec9f99b80bd1", + "prepared_review_unit_hash": "edbe5673f6f45cf9fd77f733d67148e16fbf59186a99dba32cc077132ce7b050", "component_budget": { "node_count": 2, "item_count": 2, @@ -26334,18 +26513,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related but distinct pipeline additions: one adds Promptable Concept Segmentation, the other adds Promptable Visual Segmentation. They share documentation/registry plumbing, but the model integrations, tests, and target task are different enough that they should not be deduplicated as one PR.", - "confidence": 0.96, + "summary": "Two open PRs add adjacent but distinct segmentation pipelines: one for promptable concept segmentation and one for promptable visual segmentation. They share scaffold/docs machinery, but target different tasks and model integrations, so they should not be deduplicated.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "Neither PR is a duplicate of the other; each introduces a different pipeline and different model/test coverage, so there is no canonical PR among them.", + "canonical_pr_reason": "PR #43613 is the broader, more substantial pipeline addition and covers more model/test integrations, so it is the better representative PR for this cluster.", "best_issue_reason": null, - "best_pr_reason": "PR 43613 is the broader change if you need one representative because it touches more model integrations and tests, but it is not a substitute for 43612 since it covers a different task.", + "best_pr_reason": "PR #43613 is the stronger global representative because it spans a wider set of related model backends and pipeline/docs updates.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "They only overlap on shared docs/registry plumbing. The actual features are different pipeline additions with different target tasks, different model files, and different test suites, so they are not the same underlying change and should not be merged as duplicates." + "reason": "They are related segmentation pipeline additions, but they implement different concrete tasks (concept vs visual segmentation) with different model/test targets, so they are not the same underlying change and would not naturally merge into one PR." } ] }, @@ -26375,40 +26554,40 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "a4508e6289f4006d8f6942f21c893617c65f5cbb649157c5f5cdfdea78c7abfc", + "prepared_review_unit_hash": "40bcf4102ef83ec08d7f44e1a05ad1484dd65935e2cd13a707956413c98a980f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4292, - "estimated_input_tokens": 1073, - "estimated_eval_tokens": 2402 + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4292, - "estimated_input_tokens": 1073, - "estimated_eval_tokens": 2402 + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs both touch `trainer.py`, but they solve different problems: one changes gradient-accumulation loss scaling, the other refactors signature-column filtering for custom trainers. They should not be clustered as duplicates.", - "confidence": 0.97, + "summary": "Both PRs touch `src/transformers/trainer.py` and add custom-trainer extensibility, but they address different code paths: one changes gradient-accumulation loss scaling, the other refactors signature-column filtering.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "If a representative PR is needed, 43651 is the more concrete standalone behavior change; 43663 is a separate hook/refactor for column filtering, not the same fix.", "best_issue_reason": null, - "best_pr_reason": "If a single PR must represent the cluster, #43651 is the better fit because it introduces a concrete trainer behavior change and new API hook; #43663 is a separate refactor for input-column filtering.", + "best_pr_reason": "43651 is the stronger standalone candidate in this cluster, but it should not be treated as a duplicate of 43663.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Different underlying changes and code paths: gradient accumulation loss scaling vs. signature-column filtering for PEFT/custom trainers. Shared file alone is not enough to treat them as duplicates." + "reason": "Different underlying Trainer changes: 43651 alters loss scaling during gradient accumulation, while 43663 adds an overridable signature-column helper for dataset filtering. Same file/subsystem is not enough to merge them." } ] }, @@ -26438,12 +26617,12 @@ "soft_pairs": [ "pull_request:43656|pull_request:43842" ], - "prepared_review_unit_hash": "05c1a30e51e854496be45532d4bd8fff9031455f7c66474fa87a0c6307e2cc9b", + "prepared_review_unit_hash": "5017c0ee5e6fac8fe6ea0d94726537cfd095ac8b431231feb223224a1dbc12a8", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3663, + "serialized_chars": 3662, "estimated_input_tokens": 916, "estimated_eval_tokens": 2088 }, @@ -26451,7 +26630,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3663, + "serialized_chars": 3662, "estimated_input_tokens": 916, "estimated_eval_tokens": 2088 }, @@ -26460,18 +26639,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two PRs in the same file both address TypeAdapter-related CLI typing/runtime issues, but they fix different failure modes: one is a future-annotations/type-hint NameError cleanup, the other is a pydantic-optional import problem.", - "confidence": 0.9, + "summary": "Both PRs touch the same CLI server file and involve TypeAdapter-related typing, but they fix different NameError paths: one is a broader annotation cleanup, the other is a specific pydantic-absent import/annotation issue. They should not be merged as duplicates.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43656 is the broader, more central fix for the CLI TypeAdapter NameError and is the better representative of the cluster.", + "canonical_pr_reason": "43842 is the more concrete and narrowly scoped fix: it directly targets the TypeAdapter NameError when pydantic is not installed by moving the import under TYPE_CHECKING and quoting the annotation.", "best_issue_reason": null, - "best_pr_reason": "PR #43656 is the best representative because it addresses the core CLI NameError directly; #43842 is a narrower follow-up for the pydantic-not-installed path.", + "best_pr_reason": "43842 is the best representative PR because it has a clear user-facing bug description, an explicit issue target, and a focused code change tied to one concrete failure mode.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": false, - "reason": "Reject: they share a file and symptom area, but the fixes target different concrete problems. #43656 changes annotations/imports to resolve a TypeAdapter NameError in the CLI, while #43842 specifically avoids importing TypeAdapter at runtime when pydantic is absent. They are not the same underlying bug and would not naturally merge into one PR." + "reason": "They overlap in file and theme, but fix different problems: 43656 is a broader typing/annotation change in serve.py, while 43842 specifically fixes TypeAdapter access when pydantic is missing. Not the same underlying bug." } ] }, @@ -26664,334 +26843,316 @@ "pull_request:44044|pull_request:44154", "pull_request:44024|pull_request:44076" ], - "prepared_review_unit_hash": "5fd4b3486ec01159309957f927ef6b60bb2ac957d9313ca4dd9e856236f34b73", + "prepared_review_unit_hash": "00cb590ca571dfc345b7a06f808293d4a590f180d0b5b217eb3b00c70b7f9ca1", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84505, - "estimated_input_tokens": 21127, - "estimated_eval_tokens": 42510 + "serialized_chars": 84512, + "estimated_input_tokens": 21128, + "estimated_eval_tokens": 42512 }, "budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 72, - "serialized_chars": 76937, - "estimated_input_tokens": 19235, - "estimated_eval_tokens": 38726 + "serialized_chars": 76943, + "estimated_input_tokens": 19236, + "estimated_eval_tokens": 38728 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is a broad family of model-specific refactors toward standardized output tracing. The packet does include multiple issue-linked PRs targeting #43979; the tightest duplicate candidates are within the GPT-J/CodeGen and GPT-Neo subclusters, while most other PRs are only thematically similar and not the same concrete change.", - "confidence": 0.94, - "canonical_issue_reason": null, - "canonical_pr_reason": "Several PRs explicitly target #43979, so the packet is not issue-free. Among them, 44722 is the best canonical PR because it is a concrete multi-file GPT-J/CodeGen output-tracing refactor on the shared issue-linked code path and is the strongest representative of that subcluster.", - "best_issue_reason": null, - "best_pr_reason": "44722 is the strongest global PR candidate because it most clearly represents the shared GPT-J/CodeGen output-tracing change family, with explicit issue linkage and a broad, concrete implementation rather than a narrow or unrelated model-specific variant.", + "summary": "This cluster is mostly a family of model-specific PRs migrating output handling to the new capture_outputs/can_return_tuple pattern. The tracking issue is shared, but most PRs touch different model code paths, so only a few are true near-duplicates.", + "confidence": 0.79, + "canonical_issue_reason": "Issue #43979 is the umbrella tracker referenced by many PRs here and best matches the shared output-tracing migration theme.", + "canonical_pr_reason": "PR #44066 is the best representative: it has the explicit issue link, a coherent GPT-J/CodeGen refactor, and sits at the center of the GPT-J follow-up/overlap group.", + "best_issue_reason": "#43979 is the only clear cluster-level issue and captures the common migration goal across the PRs.", + "best_pr_reason": "#44066 is the cleanest and most representative PR for this cluster; later GPT-J variants look like overlapping follow-ups rather than a better canonical anchor.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Different model families and code paths (CVT/FNet vs GPT-J); only the broad tracing theme overlaps." + "reason": "Different model families and code paths (CVT/FNet vs GPT-J); same issue theme is not enough." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different architectures and files; same standardized-output initiative, not the same concrete fix." + "reason": "Different models and unrelated forward/output implementations; not the same concrete change." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Both are GPT-J/CodeGen output-tracing refactors on the same underlying code path, so they look mergeable into one change." + "reason": "Both target the same GPT-J output-tracing refactor; 44066 includes the same GPT-J path and could plausibly absorb 44085." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "ResNet/RegNet vs EfficientNet are separate model-specific refactors, not one bug or change." + "reason": "Both are standardized-output refactors, but they touch different backbone models and distinct code paths." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "EfficientNet and GPT-J/CodeGen touch unrelated code paths; same pattern is not enough." + "reason": "EfficientNet vs GPT-J/CodeGen are different model families and not a single mergeable change." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "GPT-J/CodeGen vs EfficientNet are unrelated implementations." + "reason": "GPT-J/CodeGen versus EfficientNet: shared refactor theme only, not the same underlying code-path bug." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 and DeBERTa-v2 are different model code paths." + "reason": "MobileNetV2 and DeBERTa-v2 are unrelated implementations; no shared concrete change." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen and MGP-STR are unrelated models despite similar decorator wording." + "reason": "GPT-J/CodeGen versus MGP-STR are different code paths and model families." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J/CodeGen vs MPT are separate concrete refactors." + "reason": "GPT-J/CodeGen and MPT are separate model implementations; not a duplicate." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Both are GPT-Neo output-tracing refactors on the same model file/test path, so they appear to be the same underlying change family." + "reason": "Same GPT-Neo refactor in the same file family; 44068 looks like a later iteration of the 44018 change." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are different model implementations." + "reason": "GPT-Neo versus GPT-J/CodeGen are different model paths, so they should stay separate." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J/CodeGen and GPT-Neo are distinct code paths." + "reason": "GPT-J/CodeGen and GPT-Neo are different code paths despite the same migration theme." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet output tracing and GPT-J/CodeGen output tracing are unrelated concrete changes." + "reason": "ResNet output tracing and GPT-J/CodeGen output tracing are different model implementations." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet and GPT-J are separate model-specific refactors." + "reason": "ResNet and GPT-J are unrelated code paths." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR and GPT-J/CodeGen are unrelated models; same decorator style is not enough." + "reason": "MGP-STR and GPT-J/CodeGen are different model families and cannot be merged as one PR." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are different architectures and code paths." + "reason": "GPT-Neo and ResNet are distinct models with separate output logic." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet and MPT are unrelated concrete changes." + "reason": "ResNet and MPT are different code paths; same output-tracing idea is too broad." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert/XLM are unrelated model paths." + "reason": "GPT-Neo and Flaubert are unrelated implementations." }, { "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert/XLM are separate refactors." + "reason": "GPT-Neo and Flaubert do not share the same concrete code-path problem." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are different implementations." + "reason": "GPT-Neo versus GPT-J/CodeGen are different model paths." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT and GPT-J/CodeGen are different code paths." + "reason": "MPT and GPT-J/CodeGen are unrelated implementations." }, { "left": "pull_request:44018", "right": "pull_request:44085", "accept": false, - "reason": "GPT-Neo and GPT-J are separate model-specific refactors." + "reason": "GPT-Neo and GPT-J are different model families." }, { "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo and MGP-STR are unrelated." + "reason": "GPT-Neo and MGP-STR are separate code paths." }, { "left": "pull_request:44071", "right": "pull_request:44085", "accept": false, - "reason": "MPT and GPT-J are not the same concrete fix." + "reason": "MPT and GPT-J are different models; no shared concrete bug/change." }, { "left": "pull_request:44068", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo and MGP-STR are unrelated code paths." - }, - { - "left": "pull_request:44085", - "right": "pull_request:44086", - "accept": false, - "reason": "GPT-J and MGP-STR are different models; shared tracing language is only thematic." + "reason": "GPT-Neo and MGP-STR are unrelated." }, { "left": "pull_request:44059", "right": "pull_request:44086", "accept": false, - "reason": "GPT-2 and MGP-STR are unrelated model-specific changes." + "reason": "GPT-2 and MGP-STR are different code paths." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT and Flaubert/XLM are different model implementations." + "reason": "ViLT and Flaubert/XLM are different model families; same refactor theme is not enough." }, { "left": "pull_request:44059", "right": "pull_request:44722", "accept": false, - "reason": "GPT-2 and GPT-J/CodeGen are separate code paths." + "reason": "GPT-2 and GPT-J/CodeGen are separate model implementations." }, { "left": "pull_request:44017", "right": "pull_request:44073", "accept": false, - "reason": "SegFormer and VisualBERT are unrelated model refactors." + "reason": "Segformer and VisualBert are unrelated code paths." }, { "left": "pull_request:44017", "right": "pull_request:44161", "accept": false, - "reason": "SegFormer and LongT5 are different concrete changes." - }, - { - "left": "pull_request:43997", - "right": "pull_request:44073", - "accept": false, - "reason": "RegNet and VisualBERT are separate model-specific changes." - }, - { - "left": "pull_request:44017", - "right": "pull_request:44071", - "accept": false, - "reason": "SegFormer and MPT do not share the same concrete code path." - }, - { - "left": "pull_request:43997", - "right": "pull_request:44017", - "accept": false, - "reason": "RegNet and SegFormer are different architectures; only the broad pattern matches." + "reason": "Segformer and LongT5 are different models; not the same concrete change." }, { - "left": "pull_request:43997", - "right": "pull_request:44056", + "left": "pull_request:44056", + "right": "pull_request:44101", "accept": false, - "reason": "RegNet and MPNet are unrelated refactors." + "reason": "MPNet and Flaubert/XLM are distinct implementations." }, { "left": "pull_request:44071", "right": "pull_request:44076", "accept": false, - "reason": "MPT and ImageGPT are separate implementations." + "reason": "MPT and ImageGPT are different code paths." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet and Flaubert/XLM are unrelated model paths." + "reason": "TextNet and Flaubert/XLM are different models." }, { "left": "pull_request:44056", "right": "pull_request:44161", "accept": false, - "reason": "MPNet and LongT5 are distinct code paths." + "reason": "MPNet and LongT5 are separate code paths." }, { "left": "pull_request:43997", "right": "pull_request:44018", "accept": false, - "reason": "RegNet and GPT-Neo are unrelated concrete changes." + "reason": "RegNet and GPT-Neo are unrelated model families." }, { "left": "pull_request:43997", - "right": "pull_request:44161", + "right": "pull_request:44056", "accept": false, - "reason": "RegNet and LongT5 are different model implementations." + "reason": "RegNet and MPNet are different code paths." }, { - "left": "pull_request:44074", - "right": "pull_request:44161", + "left": "pull_request:44071", + "right": "pull_request:44129", "accept": false, - "reason": "TextNet and LongT5 are not the same bug or change." + "reason": "MPT and SpeechT5 are unrelated implementations." }, { - "left": "pull_request:44018", + "left": "pull_request:43996", "right": "pull_request:44161", "accept": false, - "reason": "GPT-Neo and LongT5 are unrelated model-specific refactors." + "reason": "CVT/FNet and LongT5 are different model paths." }, { - "left": "pull_request:44010", - "right": "pull_request:44071", + "left": "pull_request:43996", + "right": "pull_request:44017", "accept": false, - "reason": "SqueezeBERT and MPT touch different code paths." + "reason": "CVT/FNet and Segformer are separate models." }, { - "left": "pull_request:44018", - "right": "pull_request:44129", + "left": "pull_request:44010", + "right": "pull_request:44073", "accept": false, - "reason": "GPT-Neo and SpeechT5 are unrelated model-specific refactors." + "reason": "SqueezeBert and VisualBert are different code paths." }, { "left": "pull_request:44010", - "right": "pull_request:44073", + "right": "pull_request:44071", "accept": false, - "reason": "SqueezeBERT and VisualBERT are different implementations." + "reason": "SqueezeBert and MPT are unrelated implementations." }, { "left": "pull_request:44010", - "right": "pull_request:44018", + "right": "pull_request:44161", "accept": false, - "reason": "SqueezeBERT and GPT-Neo are separate changes." + "reason": "SqueezeBert and LongT5 do not share the same concrete change." }, { - "left": "pull_request:44071", - "right": "pull_request:44129", + "left": "pull_request:43997", + "right": "pull_request:44074", "accept": false, - "reason": "MPT and SpeechT5 are unrelated code paths." + "reason": "RegNet and TextNet are different backbone families." }, { - "left": "pull_request:43996", - "right": "pull_request:44161", + "left": "pull_request:44074", + "right": "pull_request:44101", "accept": false, - "reason": "CVT/FNet and LongT5 are unrelated model refactors." + "reason": "TextNet and Flaubert/XLM are unrelated model paths." }, { - "left": "pull_request:43996", - "right": "pull_request:44017", + "left": "pull_request:44056", + "right": "pull_request:44071", + "accept": false, + "reason": "MPNet and MPT are different code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44059", "accept": false, - "reason": "CVT/FNet and SegFormer are different implementations." + "reason": "GPT-Neo and GPT-2 are separate model implementations." }, { "left": "pull_request:44010", "right": "pull_request:44044", "accept": false, - "reason": "SqueezeBERT and DeBERTa-v2 are separate model-specific changes." + "reason": "SqueezeBert and DeBERTa-v2 are distinct models." }, { "left": "pull_request:44044", @@ -27003,46 +27164,58 @@ "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Both are GPT-J/CodeGen output-tracing refactors on the same underlying file set, so they are the same change family." + "reason": "Near-identical GPT-J/CodeGen output-tracing refactors on the same files; 44722 looks like a later overlapping iteration of 44066." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both touch GPT-J output tracing in the same code path, making them overlapping implementations of the same concrete refactor." + "reason": "Both touch GPT-J output tracing and appear to be overlapping versions of the same refactor, with 44722 adding the same GPT-J path plus CodeGen." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J and SpeechT5 are unrelated model paths." + "reason": "GPT-J and SpeechT5 are different model code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and ResNet/RT-DETR are different backbones; same issue target is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 are unrelated implementations." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa-v2 and SpeechT5 are separate concrete refactors." + "reason": "DeBERTa-v2 and SpeechT5 are different code paths." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa-v2 and ViLT are not the same underlying fix." + "reason": "DeBERTa-v2 and VITS are separate models and not a single duplicate fix." }, { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "FocalNet and ImageGPT are different models; only the decorator pattern is shared." + "reason": "FocalNet and ImageGPT are different model families; shared output-tracing pattern is too broad." } ] }, "evaluator_result": { - "accept": false, - "feedback": "Reject: the output is not fully grounded in the packet. The canonical_pr_reason says \"No issue artifacts are present,\" but the packet shows multiple PRs with explicit_issue_links/targets, including 44722 and 44085 (target 43979) and others. That is a factual overstatement. The soft-edge verdicts are mostly conservative, but the summary and canonical reasoning need to reflect the presence of issue-linked PRs rather than denying them." + "accept": true, + "feedback": "The summary is grounded in the packet: the shared tracker issue #43979 is clearly central, and the canonical PR choice (#44066) is supported by its explicit issue link and its GPT-J/CodeGen overlap with later follow-ups. The soft-edge verdicts are mostly conservative, and the accepted pairs are the ones that are visibly same-family, same-file, or explicit overlap cases." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -27074,64 +27247,64 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "073ae63e7592770de77b5e2f716609bc48878dab1395f08d0902074b304f6d99", + "prepared_review_unit_hash": "d286beac73f431011dbdc8c028d3609a9ffc5ad9cbd5a3ec35211edb75255083", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84505, - "estimated_input_tokens": 21127, - "estimated_eval_tokens": 42510 + "serialized_chars": 84512, + "estimated_input_tokens": 21128, + "estimated_eval_tokens": 42512 }, "budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 5, - "serialized_chars": 13068, - "estimated_input_tokens": 3267, - "estimated_eval_tokens": 6790 + "serialized_chars": 13069, + "estimated_input_tokens": 3268, + "estimated_eval_tokens": 6792 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These PRs are all model-specific refactors for standardized output tracing/capture, but they touch different architectures and code paths, so they are not duplicates of one another.", - "confidence": 0.96, + "summary": "This cluster is a set of model-specific refactors to standardized output tracing/capture_outputs. They are stylistically similar, but they target different model implementations and code paths, so they should not be merged as duplicates.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "No PR here is a true duplicate of another; if one representative is needed, 44073 is a clean example of the shared output-tracing refactor pattern, but it remains specific to VisualBert.", + "canonical_pr_reason": "No single PR is a true duplicate of the others; if one must represent the cluster, 44129 is the most complete anchor because it has an explicit issue target and a concrete, self-contained SpeechT5 output-tracing refactor.", "best_issue_reason": null, - "best_pr_reason": "44073 is the most representative of the cluster\u2019s common theme: a straightforward model-level migration to capture_outputs/can_return_tuple with explicit output-recording hooks.", + "best_pr_reason": "44129 is the strongest representative to anchor discussion because it includes an explicit linked issue and a substantive output-capture change, while the other PRs are parallel model-specific refactors.", "soft_edge_verdicts": [ { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Same broad refactor theme, but different models and files (VisualBert vs TextNet); not the same underlying change or code path." + "reason": "Both are output-tracing refactors, but they apply to different models (VisualBert vs TextNet) and different files/code paths; same pattern, not the same change." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "Both are output-tracing migrations, but they target different architectures and internal modules, so they are not mergeable as one PR." + "reason": "MPNet vs VisualBert are distinct model implementations; these are parallel decorator refactors rather than one underlying bug or mergeable single PR." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "Different model implementations and distinct refactor scopes; shared decorator usage alone is not enough for duplicate status." + "reason": "Different models and different forward/output-capture behavior; the similarity is only the standardized refactor pattern." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert are unrelated code paths; these are separate model migrations, not the same fix." + "reason": "RegNet and SqueezeBert are unrelated code paths; both migrate to output tracing, but they are not the same concrete fix." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert output-collection refactor and SpeechT5 output tracing are separate model-specific changes; SpeechT5 also has an explicit issue target and a different functional area." + "reason": "SqueezeBert and SpeechT5 have different model-specific output handling and touched code; they are thematically similar refactors, not duplicates." } ] }, @@ -27211,274 +27384,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "2f6a7d3f917139a0dd04e1e944ea1b06b50c157291955c4b00b181809c9630aa", + "prepared_review_unit_hash": "f5568d0a3af1f75c605ac987fb3eaedfb5665c7b6549a2643d5a2cfe9bc9479f", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40177, - "estimated_input_tokens": 10045, - "estimated_eval_tokens": 20346 + "serialized_chars": 40184, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40177, - "estimated_input_tokens": 10045, - "estimated_eval_tokens": 20346 + "serialized_chars": 40184, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs all share a surface theme around output-tracing refactors, but they target different model implementations and code paths, so they should not be deduplicated against each other.", - "confidence": 0.95, + "summary": "This cluster is a set of similarly named output-tracing refactors across unrelated model implementations, not duplicate fixes for one underlying bug. None of the soft edges look safe to merge as duplicates.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR as a duplicate anchor: the changes are model-specific refactors across unrelated forward paths, not one shared bug or fix.", + "canonical_pr_reason": "No single canonical PR: the items target different model codepaths, so they are only thematically similar.", "best_issue_reason": null, - "best_pr_reason": "PR 44003 is the strongest representative of the theme because it spans two closely related Mamba-family files and adds the tracing/capture plumbing, but it still is not a duplicate of the other model-specific PRs.", + "best_pr_reason": "PR 44025 is the strongest representative because it updates two closely related depth-anything variants and has the broadest coherent scope in the set.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Different model code paths (timm_backbone vs upernet); same refactor theme, not the same change." + "reason": "Same refactor theme, but different model codepaths (`timm_backbone` vs `upernet`)." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different architectures and forward implementations; only share a broad output-tracing pattern." + "reason": "Both are output-tracing refactors, but they target unrelated models and cannot be merged as one fix." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Unrelated model code paths (timm_backbone vs univnet)." + "reason": "Different architectures and codepaths; only the high-level refactor theme matches." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "Different families and files; both refactor output handling, but not one concrete underlying fix." + "reason": "Different model families (`timm_backbone` vs `falcon_mamba`/`mamba`); not the same underlying change." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "timm_backbone and rwkv are unrelated implementations; no shared bug or merged PR candidate." + "reason": "Unrelated model implementations; same pattern, not the same bug or change." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "MobileNetV1 vs UnivNet are unrelated model paths." + "reason": "Different model families; this is a shared refactor pattern, not a duplicate fix." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Different models and behaviors; only similar output-tracing cleanup." + "reason": "Separate codepaths in different models; no single concrete change in common." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "MobileNetV1 and Depth Anything are separate model implementations." + "reason": "Both touch output tracing, but on different model stacks and files." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "No shared concrete code path; different seq2seq wrapper vs image model." + "reason": "Different encoder-decoder vs vision model codepaths; not mergeable as one PR." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "MobileNetV1 and Speech Encoder-Decoder are unrelated." + "reason": "Same theme, but distinct model implementations and fixes." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Vision-text dual encoder vs UnivNet; same refactor wording, different code paths." + "reason": "These are unrelated model implementations; similarity is only in the refactor pattern." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different model families; not the same underlying bug or change." + "reason": "Different model codepaths; not the same concrete bug/change." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "Vision-text dual encoder and Mamba family do not share a concrete fix path." + "reason": "Separate models with no shared fix beyond the generic output-tracing refactor." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Different models and call stacks; only a broad tracing refactor similarity." + "reason": "Different model families and files; no single duplicate change." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "Unrelated model implementations; not mergeable as one PR." + "reason": "Only the refactor motif matches; the underlying implementations are unrelated." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "UnivNet and Mamba are different code paths with different output semantics." + "reason": "Different model stacks and concrete codepaths." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "Different model architectures and forward signatures." + "reason": "Different model families; not the same underlying fix." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "UperNet vs RWKV are unrelated model paths." + "reason": "Separate models and implementations; no shared concrete bug." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "Different model families; no single concrete bug/fix shared." + "reason": "Different model codepaths, despite similar output-tracing work." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "Mamba/Falcon-Mamba vs SuperPoint are unrelated implementations." + "reason": "Different model families and implementation details; not duplicate PRs." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "Mamba-family output tracing and CodeGen output tracing are separate code paths." + "reason": "These refactors touch different concrete model logic (`mamba` vs `codegen`)." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "Timm backbone and CodeGen are unrelated; only superficial refactor similarity." + "reason": "Different models and change sets; same pattern only at a very high level." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "UnivNet and CodeGen do not fix the same concrete problem." + "reason": "Unrelated codepaths; not the same underlying bug or change." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "UperNet and CodeGen are different models and output paths." + "reason": "Different model implementations; no safe duplicate merge." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "Different architectures; not the same underlying change." + "reason": "Same refactor category, but different model-specific fixes." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Depth Anything and Vision Encoder-Decoder are different models and wrappers." + "reason": "Different model families (`depth_anything` vs `vision_encoder_decoder`); not mergeable as one PR." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Different model families; no shared code-path bug." + "reason": "Different codepaths and model stacks; only the refactor theme is shared." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "Depth Anything and SuperPoint are unrelated vision models." + "reason": "Separate models with no shared concrete bug or implementation." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Different implementations and output shapes; not a duplicate." + "reason": "Different models; too broad to treat as duplicate changes." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Depth Anything and DPR solve different tasks and code paths." + "reason": "Different architectures and files; not the same underlying change." }, { "left": "pull_request:44026", "right": "pull_request:44029", "accept": false, - "reason": "Vision Encoder-Decoder and RWKV are unrelated models." + "reason": "Different model families and codepaths; similarity is only thematic." }, { "left": "pull_request:44026", "right": "pull_request:44027", "accept": false, - "reason": "Different seq2seq wrappers and output plumbing; not one fix." + "reason": "Both are encoder-decoder style refactors, but they target different model implementations." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "Vision Encoder-Decoder and SuperPoint are unrelated." + "reason": "Unrelated models and concrete fixes." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "Different model families and forward contracts." + "reason": "Different model stacks and no shared bug report or codepath." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "Speech Encoder-Decoder and SuperPoint are separate code paths." + "reason": "Different model implementations; not the same concrete PR." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "Speech Encoder-Decoder and RWKV are unrelated." + "reason": "Separate codepaths in unrelated models." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations; no shared concrete bug." + "reason": "Different model families and fix scopes." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "SuperPoint and RWKV are unrelated models." + "reason": "Different concrete implementations; no shared duplicate change." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "SuperPoint and DPR do not share the same fix target." + "reason": "Unrelated model codepaths, despite similar refactor wording." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "RWKV and DPR are different architectures with different output semantics." + "reason": "Different model families and changes; not a duplicate fix." } ] }, @@ -27508,7 +27681,7 @@ "soft_pairs": [ "pull_request:44662|pull_request:45401" ], - "prepared_review_unit_hash": "aa59e2402ec558cf391a76f5f4f05f7038317e3da31b4df398b96d5fd597b81a", + "prepared_review_unit_hash": "b5285192e7f931d08fe358e603c4f6d6a7e0991dfd63db2e82c6cd4d156384d6", "component_budget": { "node_count": 2, "item_count": 2, @@ -27530,18 +27703,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are unrelated model-addition PRs: one introduces PenguinVL, the other adds Voxtral TTS support. They share only generic transformer registry/docs touchpoints, not the same underlying change.", - "confidence": 0.98, + "summary": "These are two unrelated model-addition PRs that only share generic Transformers integration files; they do not appear to fix the same bug or implement the same change.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:44662 is the more canonical representative because it is a complete model implementation with extensive code, docs, and tests, and it is not draft.", + "canonical_pr_reason": "PR 45401 is the more focused, explicitly scoped Voxtral TTS implementation with a linked tracking issue; PR 44662 is a separate PenguinVL model addition.", "best_issue_reason": null, - "best_pr_reason": "pull_request:44662 is the best standalone representative of the cluster for triage purposes because it is the broader, more mature implementation PR; however, it is not a duplicate of the other PR.", + "best_pr_reason": "45401 is the best representative of the cluster only in the sense that it is a self-contained model-support PR; 44662 is clearly a different model and change set.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Different features and code paths: PenguinVL implementation vs Voxtral TTS support. Shared auto/docs files are generic registration changes, not evidence of the same bug or change." + "reason": "Reject: PenguinVL implementation vs Voxtral TTS support are different models with different code paths and tests; overlapping registry/docs files are generic boilerplate, not evidence of the same underlying change." } ] }, @@ -27571,7 +27744,7 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "2ad26328afc35b1e4cf550ebc9f6f01e068a84a9761e90f07d71808b91b547f3", + "prepared_review_unit_hash": "c1366eaa65a9610d3632ebd5f0bb09304a014af3ffedbc51c99a8231d0a34a26", "component_budget": { "node_count": 2, "item_count": 2, @@ -27593,18 +27766,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are related to module/kernel fusion, but they are not duplicates: one introduces a generic module-fusion API, while the other adds a concrete n-to-1 kernel fusion integration via KernelConfig and model-loading changes.", - "confidence": 0.91, + "summary": "Two related pull requests, but they are not duplicates: one introduces the core Module Fusion API, while the other adds a specific n-to-1 kernel fusion workflow built on top of that API.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45363 is the more complete, user-facing implementation: it wires kernel fusion through modeling/utils/integration code and adds an example, whereas PR 44979 is mainly a foundational module-fusion API and tests.", + "canonical_pr_reason": "PR #44979 is the broader foundational change: it adds the new `module_fusion.py` API and tests, which is the base abstraction other fusion features depend on.", "best_issue_reason": null, - "best_pr_reason": "PR 45363 is the best representative of the cluster because it implements a specific end-to-end feature path rather than just an abstraction layer.", + "best_pr_reason": "PR #44979 is the best cluster representative because it is the general reusable API change; #45363 is a narrower follow-on implementation using that API.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "Related area, but not the same concrete change. 44979 adds a generic module-fusion API in a new file, while 45363 implements n-to-1 kernel fusion through KernelConfig, model loading, and integration hooks. They could coexist, not be merged as one duplicate PR." + "reason": "They address different levels of the stack: #44979 adds the generic module-fusion API, while #45363 wires a specific `KernelConfig`-based n-to-1 kernel fusion path and example. Related, but not the same concrete change and not something that would plausibly be merged as one PR." } ] }, @@ -27634,7 +27807,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "0a94f6f54fd8cd008b9a72410e944a2bfa31ec0b161c55af95dc193dab343a08", + "prepared_review_unit_hash": "88597779aa9af70ece8bb78010f1f716da9628895b06b7ec93d529048e0d3461", "component_budget": { "node_count": 2, "item_count": 2, @@ -27656,18 +27829,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same AttributeError in `_patch_mistral_regex` for the Mistral regex patch, and the second PR is a fuller version with an added test. This is a duplicate fix cluster centered on the same code path.", + "summary": "Both PRs fix the same AttributeError in `_patch_mistral_regex` by switching from `tokenizer.backend_tokenizer.pre_tokenizer` to `tokenizer.pre_tokenizer`; 45317 is the more complete version because it adds test coverage.", "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45317 is the better representative because it targets the same fix but adds test coverage and a clearer problem statement, making it the more complete implementation of the bug fix.", + "canonical_pr_reason": "PR 45317 is the better canonical because it addresses the same bug as 45086 but also includes tests, making it the more complete change for the same code path.", "best_issue_reason": null, - "best_pr_reason": "PR 45317 is the strongest overall artifact for the cluster: same underlying bug, same file, and it includes tests, so it better captures the final intended change.", + "best_pr_reason": "PR 45317 is the best PR because it contains the same fix plus validation, so it is the strongest representative of this duplicate cluster.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying AttributeError fix in `_patch_mistral_regex` on the same code path; the second PR appears to be a more complete iteration of the first rather than a different change." + "reason": "Same underlying bug and same patch hunk in the same file; 45317 is essentially 45086 plus tests, so these are duplicate PRs for the same fix." } ] }, @@ -27697,12 +27870,12 @@ "soft_pairs": [ "pull_request:45115|pull_request:45569" ], - "prepared_review_unit_hash": "8559afb3e5cecb9fac5e8bf65e18fd0f67188d31062b7841e482b8d8439bc9eb", + "prepared_review_unit_hash": "51ef619e9ea5e5ecb8a90dd01c9c5f92244f8ddc442be054c262e89c03c723c9", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6066, + "serialized_chars": 6065, "estimated_input_tokens": 1517, "estimated_eval_tokens": 3290 }, @@ -27710,7 +27883,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6066, + "serialized_chars": 6065, "estimated_input_tokens": 1517, "estimated_eval_tokens": 3290 }, @@ -27719,18 +27892,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open NemotronH pull requests overlap on model internals, but they are not the same change: one is a smaller refactor/conversion-mapping update, the other is a broader split into dedicated dense/sparse model packages with docs and tests.", - "confidence": 0.82, + "summary": "These are related NemotronH PRs, but they target different changes: one is a refactor to make NemotronH inherit GraniteMoeHybrid and update conversion mapping, while the other is a broader model split into NemotronH Dense/Sparse with docs and auto-mapping updates. Overlap in shared files is not enough to treat them as duplicates.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45569 is the better canonical representative because it covers the larger, more complete architecture change: dedicated NemotronHDense/NemotronHSparse modules, auto-mapping updates, docs, and tests. PR #45115 is a narrower precursor refactor focused on inheritance/conversion naming.", + "canonical_pr_reason": "PR 45569 is the better canonical cluster representative because it is broader and more user-visible, covering the model split, docs, and auto-mapping changes rather than just the narrower inheritance/conversion refactor.", "best_issue_reason": null, - "best_pr_reason": "PR #45569 is the best single PR in this cluster since it most clearly defines the final intended model packaging and user-facing behavior. PR #45115 appears to be an earlier internal refactor rather than the main functional change.", + "best_pr_reason": "PR 45569 is the best single PR for this cluster because it encompasses the more complete NemotronH architecture update and superseding model-layout changes.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "Shared NemotronH files alone are not enough: #45115 is a conversion/refactor to GraniteMoeHybrid-compatible naming, while #45569 splits NemotronH into dense/sparse variants and updates docs/tests. They are related but not the same underlying fix and would not be merged into one PR." + "reason": "Reject: they touch the same NemotronH codebase, but 45115 focuses on GraniteMoeHybrid-compatible inheritance and checkpoint conversion renames, while 45569 adds NemotronH Dense/Sparse modules, docs, and auto-model wiring. Different concrete changes, not one mergeable fix." } ] }, @@ -27760,7 +27933,7 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "d5689837f48a2fdac94e0178a62d9124e69bcddbb2f6973f9b1f792ac5d255f7", + "prepared_review_unit_hash": "eef3625725490ba020a69a4131204754716c55c4490e74272144135a7347b79b", "component_budget": { "node_count": 2, "item_count": 2, @@ -27782,18 +27955,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate typing-related pull requests, not duplicates. PR 45415 is a broad repo-wide type-checking sweep across many files, while PR 45425 makes a more focused internal typing refactor around `_typing.py`, `modeling_utils`, PEFT-related protocols, and type-checking helpers.", - "confidence": 0.88, + "summary": "Both pull requests are typing-related, but they address different changes: PR 45415 is a broad repo-wide type-checking pass across many `src/transformers/*.py` files, while PR 45425 introduces new typing protocols and targeted fixes around `modeling_utils`, PEFT, and flash-attention helpers. They overlap in theme and a few files, but not the same concrete change.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45425 is the better canonical representative for this cluster because it is more focused and has a clearer concrete change-set centered on shared typing abstractions, whereas PR 45415 is a broad type-checking pass across many modules.", + "canonical_pr_reason": "PR 45415 is the broader umbrella change: it touches many files to make `src/transformers/*py` type-check cleanly, so it best represents the cluster\u2019s main typing effort.", "best_issue_reason": null, - "best_pr_reason": "PR 45425 is the best standalone PR in the pair: it targets a specific typing infrastructure change and is easier to evaluate independently. PR 45415 is broader and more of a sweeping cleanup than a single change.", + "best_pr_reason": "PR 45415 is the best single representative because it is the most comprehensive type-checking pass and captures the cluster\u2019s dominant scope better than the narrower follow-up typing PR.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Both are typing-related, but they address different scopes and concrete changes. PR 45415 is a broad type-checking sweep across many files; PR 45425 introduces specific typing protocols and integrations. They would not plausibly be merged as one duplicate PR." + "reason": "Reject: both are typing cleanups, but 45415 is a broad type-checking sweep while 45425 adds specific typing protocols and targeted fixes; they are not the same underlying bug/change and would not naturally merge into one PR." } ] }, @@ -27823,12 +27996,12 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "6fb3270b9fa6bcb68fde994b78815f4f1728d7f99388d339756cd32522097f2e", + "prepared_review_unit_hash": "7a5d9b49c516be318fec15776f73c20e9daedc36635711a4d47e21cac0d7fd76", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4507, + "serialized_chars": 4508, "estimated_input_tokens": 1127, "estimated_eval_tokens": 2510 }, @@ -27836,7 +28009,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4507, + "serialized_chars": 4508, "estimated_input_tokens": 1127, "estimated_eval_tokens": 2510 }, @@ -27845,18 +28018,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs both touch expert-parallel loading, but they fix different failure modes: #45548 targets EP + DeepSpeed ZeRO-3 loading via accelerate launch and serialization of distributed_config, while #45662 targets EP + FSDP2 where experts are silently overwritten by rank-0 broadcast and adds post-shard wrapping in the loader. They are related, but not the same underlying bug/change.", - "confidence": 0.95, + "summary": "Both PRs are about expert parallelism and share `modeling_utils.py`, but they fix different concrete failures: one is about preserving `distributed_config` / `has_ep` for DeepSpeed ZeRO-3 loading via accelerate, the other about preventing expert parameters from being overwritten by rank-0 broadcast in FSDP2. They are related but not duplicates.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45662 is the stronger representative of the cluster because it addresses a concrete loader-path corruption bug with broader code-path changes; #45548 is a narrower fix for a different loading/serialization issue.", + "canonical_pr_reason": "No single canonical PR: the two patches address different distributed-loading bugs, so neither subsumes the other.", "best_issue_reason": null, - "best_pr_reason": "PR #45662 is the best candidate to keep as the cluster representative since it fixes a specific post-shard overwrite problem in the model-loading path and is more directly aligned with the shared EP loading theme.", + "best_pr_reason": "PR 45662 is the stronger representative for the cluster because it has the broader, more explicit distributed-state fix path and clearer failure mode, though it is not a duplicate of 45548.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Different concrete bugs: #45548 is about preserving distributed_config and enabling EP+ZeRO-3 loading, while #45662 fixes expert parameters being overwritten after rank-0 broadcast during FSDP2 loading. Shared subsystem only, not the same change." + "reason": "Both involve EP/loading and `modeling_utils.py`, but 45548 fixes `distributed_config` serialization / `has_ep` for ZeRO-3 loading, while 45662 fixes expert parameters being overwritten during FSDP2 post-shard handling. Different root causes and different code paths." } ] }, @@ -27886,40 +28059,40 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "fc4638da4fcf0997139222ece914d679b010bc96ff0b7323d85c78e3bd862afb", + "prepared_review_unit_hash": "972e221afea185d36a74a392ea27d770f12b5c91569b9b1caa40ebf154fc8658", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4673, - "estimated_input_tokens": 1169, - "estimated_eval_tokens": 2594 + "serialized_chars": 4672, + "estimated_input_tokens": 1168, + "estimated_eval_tokens": 2592 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4673, - "estimated_input_tokens": 1169, - "estimated_eval_tokens": 2594 + "serialized_chars": 4672, + "estimated_input_tokens": 1168, + "estimated_eval_tokens": 2592 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same underlying xdist collision bug for captured_info logs under the same tracked issue; PR 45645 is the broader, more complete fix and subsumes 45639.", - "confidence": 0.96, + "summary": "Both PRs target the same xdist-related captured_info/debug-log collision problem in `testing_utils`; PR #45645 appears to be the broader, more complete version because it adds the CI workflow change needed to consume multiple per-worker files.", + "confidence": 0.81, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45645 is the better canonical PR because it includes the core xdist-safe logging fix plus the CI workflow update to collect multiple captured_info files, making it the more complete implementation of the shared bug fix.", + "canonical_pr_reason": "PR #45645 is the better canonical PR because it covers the core xdist-safe logging fix and also updates the workflow to aggregate the new per-worker artifacts, making it the more complete end-to-end change.", "best_issue_reason": null, - "best_pr_reason": "PR 45645 is the best overall PR for this cluster since it addresses the same concrete collision problem as 45639 and also preserves CI debug-log collection, so it is the more comprehensive change.", + "best_pr_reason": "PR #45645 is the strongest overall candidate: it addresses the bug in the producer and the downstream CI log collection path, so it is the most complete fix for the underlying xdist artifact collision issue.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Same concrete bug: captured_info output collisions under pytest-xdist. The shared target and overlapping testing_utils changes indicate they could be merged into one PR, with 45645 adding workflow support on top." + "reason": "Both PRs address the same captured_info/xdist collision problem and touch the same fix path in `testing_utils`; #45645 is essentially the broader superset that adds workflow handling, so they are close enough to treat as the same underlying change." } ] }, @@ -28015,7 +28188,7 @@ "soft_pairs": [ "pull_request:46126|pull_request:46269" ], - "prepared_review_unit_hash": "31d4e5a89f0809c3775a840a32e6db702fe9ff0cf06bdc9d0fe8c94606dfaece", + "prepared_review_unit_hash": "3e41010c2264ff8682ca30d131e9807faae7e5fe30cbe15f43c4473c7ef46c0b", "component_budget": { "node_count": 2, "item_count": 2, @@ -28037,18 +28210,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are not duplicates. PR #46126 is a focused EP/trainer integration change around router/tensor-parallel behavior, while PR #46269 is a broad distributed branch base PR touching many unrelated areas. Shared filenames alone are not enough to merge them.", - "confidence": 0.91, + "summary": "These are not duplicates: one PR is a focused EP/trainer integration change, while the other is a broad distributed-branch base touching many unrelated distributed subsystems. Shared filenames alone are not enough to treat them as the same underlying change.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46126 is the better canonical PR because it is the more specific, feature-scoped change with a clear concrete code path (EP routing and trainer integration).", + "canonical_pr_reason": "PR #46126 is the more specific, self-contained change with a clear concrete purpose (EP + trainer integration) and is a better representative of the narrower topic in this cluster.", "best_issue_reason": null, - "best_pr_reason": "PR #46126 is the best representative of the cluster since it is narrowly about the underlying EP/trainer functionality, unlike the broad base-branch PR.", + "best_pr_reason": "PR #46126 is the best PR candidate because it is the more focused implementation-level change; PR #46269 is too broad and platform-like to serve as a duplicate representative.", "soft_edge_verdicts": [ { "left": "pull_request:46126", "right": "pull_request:46269", "accept": false, - "reason": "Reject: the overlap is only superficial. #46126 targets EP router masking and trainer integration, while #46269 is a large base branch PR spanning many distributed subsystems. They do not look like the same underlying fix and would not plausibly be merged as one PR." + "reason": "Different scope and intent: #46126 adds EP routing/trainer integration around tensor-parallel/MoE code paths, while #46269 is a large distributed-branch base touching many files across FSDP, integrations, modeling, and CI. They do not look like the same concrete fix or change." } ] }, @@ -28116,220 +28289,220 @@ "pull_request:46504|pull_request:46716", "pull_request:46432|pull_request:46716" ], - "prepared_review_unit_hash": "b18bb471e46d042a703ea442a6558658dd8f61c0458567d9f92dcce00750f2b2", + "prepared_review_unit_hash": "101412e7166dcfecb21581b7cc706b80c1e95de805f8ec799ba77005b6cee547", "component_budget": { "node_count": 10, "item_count": 10, "soft_pair_count": 31, - "serialized_chars": 26686, - "estimated_input_tokens": 6672, - "estimated_eval_tokens": 13600 + "serialized_chars": 26681, + "estimated_input_tokens": 6671, + "estimated_eval_tokens": 13598 }, "budget": { "node_count": 10, "item_count": 10, "soft_pair_count": 31, - "serialized_chars": 26686, - "estimated_input_tokens": 6672, - "estimated_eval_tokens": 13600 + "serialized_chars": 26681, + "estimated_input_tokens": 6671, + "estimated_eval_tokens": 13598 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are mostly small workflow maintenance changes, but they split into unrelated edits across different files and even within .github/workflows/pr-ci-security-gate.yml the changes address different concerns. I do not see any true duplicate PRs here.", - "confidence": 0.92, + "summary": "These are mostly unrelated workflow-maintenance PRs. The only dense area is pr-ci-security-gate.yml, but those PRs make different follow-up edits rather than the same concrete fix, so I would not deduplicate any pair.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No true duplicate PR stands out. If a representative is needed, PR 46554 is the closest because it makes the main functional security-gate logic change; the others are smaller or distinct follow-ups.", "best_issue_reason": null, - "best_pr_reason": "PR 46554 is the strongest representative of the cluster\u2019s most substantive workflow/security change: it adds checkout-based diff inspection and explicit failure paths for workflow/CI-infra changes. Even so, it is not a duplicate of the other PRs.", + "best_pr_reason": "PR 46554 is the strongest standalone PR: it adds actual security-gate behavior (checkout and changed-file checks), whereas the others are narrower workflow tweaks, permissions changes, or comments.", "soft_edge_verdicts": [ { "left": "pull_request:46138", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and unrelated fixes: stdin handling in self-comment CI vs security-gate logic." + "reason": "Different workflows and different fixes; same broad CI area only." }, { "left": "pull_request:46138", "right": "pull_request:46558", "accept": false, - "reason": "Different workflow files and unrelated changes; not the same bug or code path." + "reason": "Unrelated workflow update versus security-gate permissions tweak." }, { "left": "pull_request:46138", "right": "pull_request:46563", "accept": false, - "reason": "self-comment CI command invocation vs a reusable-workflow note in pr-ci-security-gate; unrelated." + "reason": "Command input handling vs. explanatory note in another workflow." }, { "left": "pull_request:46138", "right": "pull_request:46716", "accept": false, - "reason": "Different workflow behavior entirely; one updates CI comment handling, the other changes secrets passing in a security gate." + "reason": "Different workflow changes; no shared concrete bug." }, { "left": "pull_request:46138", "right": "pull_request:46504", "accept": false, - "reason": "Unrelated workflow edits in different files: stdin piping vs commit-message grep behavior." + "reason": "Separate shell-quoting fixes in different workflows." }, { "left": "pull_request:46432", "right": "pull_request:46503", "accept": false, - "reason": "Different workflow files and unrelated runner/benchmark changes." + "reason": "SSH runner hardening vs. benchmark trigger logic; distinct changes." }, { "left": "pull_request:46432", "right": "pull_request:46501", "accept": false, - "reason": "ssh-runner hardening is unrelated to anti-slop permissions placement." + "reason": "Different workflows and different problems." }, { "left": "pull_request:46432", "right": "pull_request:46502", "accept": false, - "reason": "Different workflows and different functionality; not the same underlying issue." + "reason": "Unrelated workflow edits; no same code-path issue." }, { "left": "pull_request:46432", "right": "pull_request:46554", "accept": false, - "reason": "Unrelated workflows: SSH runner hardening vs security-gate file scanning." + "reason": "Security hardening in SSH runner vs. PR-CI gate logic." }, { "left": "pull_request:46432", "right": "pull_request:46558", "accept": false, - "reason": "Different workflow paths and fixes; no shared concrete bug." + "reason": "Different workflow file and different fix scope." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46563", + "accept": false, + "reason": "No shared underlying bug; one is SSH hardening, the other is a note." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46716", + "accept": false, + "reason": "Distinct SSH workflow vs. PR CI security gate change." }, { "left": "pull_request:46501", "right": "pull_request:46503", "accept": false, - "reason": "Different workflows: anti-slop permissions vs benchmark trigger conditions." + "reason": "Different workflows and triggers; not the same change." }, { "left": "pull_request:46501", "right": "pull_request:46502", "accept": false, - "reason": "Different workflow files and unrelated updates." + "reason": "Separate CI workflows; no common concrete bug." }, { "left": "pull_request:46501", "right": "pull_request:46504", "accept": false, - "reason": "anti-slop permissions change is unrelated to build-ci-docker-images command handling." + "reason": "Permissions move in anti-slop vs. shell input quoting fix." }, { "left": "pull_request:46501", "right": "pull_request:46554", "accept": false, - "reason": "Different files and concerns; anti-slop auth scope vs PR CI security gate logic." + "reason": "Different workflow logic and distinct intent." }, { "left": "pull_request:46502", "right": "pull_request:46503", "accept": false, - "reason": "Different workflow files and unrelated reviewer-assignment vs benchmark trigger changes." + "reason": "Base-ref checkout fix vs. benchmark schedule trigger." }, { "left": "pull_request:46502", "right": "pull_request:46504", "accept": false, - "reason": "assign-reviewers checkout ref change is unrelated to build-ci-docker-images commit-message parsing." + "reason": "Different workflows and separate fixes." }, { "left": "pull_request:46502", "right": "pull_request:46554", "accept": false, - "reason": "Different workflows and unrelated fixes." + "reason": "Assign-reviewers checkout tweak is unrelated to PR-CI gate logic." }, { "left": "pull_request:46503", "right": "pull_request:46504", "accept": false, - "reason": "benchmark_v2 trigger update and build-ci-docker-images parsing change are different changes." + "reason": "Benchmark dispatch support vs. docker-image command quoting." }, { "left": "pull_request:46503", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and unrelated logic changes." + "reason": "Benchmark trigger update is unrelated to security-gate checks." }, { "left": "pull_request:46504", "right": "pull_request:46554", "accept": false, - "reason": "Different code paths: build-ci-docker-images commit parsing vs PR CI security gating." + "reason": "Same CI subsystem, but different workflows and different bugs." }, { "left": "pull_request:46504", "right": "pull_request:46558", "accept": false, - "reason": "Different workflow files and unrelated fixes." - }, - { - "left": "pull_request:46554", - "right": "pull_request:46558", - "accept": false, - "reason": "Same file but different concrete changes: file-diff security checks vs permissions/approval gating; not mergeable as one fix." + "reason": "Shell quoting fix vs. security-gate permissions/condition change." }, { - "left": "pull_request:46554", + "left": "pull_request:46504", "right": "pull_request:46563", "accept": false, - "reason": "Same workflow file but different concerns; one changes security checks, the other adds explanatory comments for a reusable workflow." + "reason": "Different workflow files and change purposes." }, { - "left": "pull_request:46554", + "left": "pull_request:46504", "right": "pull_request:46716", "accept": false, - "reason": "Same workflow file but distinct changes: security checks vs removing inherited secrets." + "reason": "Docker-image workflow fix vs. PR-CI secrets handling." }, { - "left": "pull_request:46558", - "right": "pull_request:46563", + "left": "pull_request:46554", + "right": "pull_request:46558", "accept": false, - "reason": "Same file but different changes to different parts of the job; not the same underlying bug." + "reason": "Same workflow file, but different follow-up changes, not one bug." }, { - "left": "pull_request:46558", - "right": "pull_request:46716", + "left": "pull_request:46554", + "right": "pull_request:46563", "accept": false, - "reason": "Different security-gate changes: permissions/approval flow vs secrets inheritance." + "reason": "Actual gate logic vs. a documentation comment." }, { - "left": "pull_request:46563", + "left": "pull_request:46554", "right": "pull_request:46716", "accept": false, - "reason": "Comments/documentation-style note vs secrets configuration change; not the same fix." - }, - { - "left": "pull_request:46504", - "right": "pull_request:46563", - "accept": false, - "reason": "Different workflows and unrelated fixes." + "reason": "Different security-gate aspects: file checks vs. secrets passing." }, { - "left": "pull_request:46432", + "left": "pull_request:46558", "right": "pull_request:46563", "accept": false, - "reason": "Different workflow files and unrelated changes." + "reason": "Permissions/flow control change versus comment-only update." }, { - "left": "pull_request:46504", + "left": "pull_request:46558", "right": "pull_request:46716", "accept": false, - "reason": "Different workflow files and unrelated fixes." + "reason": "Different fixes to the same file, but not the same concrete problem." }, { - "left": "pull_request:46432", + "left": "pull_request:46563", "right": "pull_request:46716", "accept": false, - "reason": "Different workflows and unrelated security/runner concerns." + "reason": "Comment-only clarification vs. secrets behavior change." } ] }, @@ -28422,7 +28595,7 @@ "soft_pairs": [ "pull_request:46310|pull_request:47162" ], - "prepared_review_unit_hash": "76afe1a81f5e12956e374d9456e70b620a691600b54223d33cec478cce35625e", + "prepared_review_unit_hash": "3f601733998f19a9ead4753c4aa69f0772a5f37164126243d89f03074f921d36", "component_budget": { "node_count": 2, "item_count": 2, @@ -28444,18 +28617,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests fix the same off-by-one bug in MoonshineStreaming\u2019s sliding_window_mask_function right-window boundary; PR 47162 is a broader, more complete version that also adds the same Moonshine fix plus Gemma4 coverage and tests.", + "summary": "Both pull requests address the same off-by-one bug in the sliding-window right mask for Moonshine Streaming; 47162 is the broader, more complete fix and also includes tests plus a related Gemma4 change with the same mask correction.", "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 47162 is the better canonical representative because it contains the same MoonshineStreaming fix as 46310, adds tests, and appears to be the more complete patch.", + "canonical_pr_reason": "PR 47162 is the better canonical PR because it applies the same right-window off-by-one fix in Moonshine Streaming, adds test coverage, and is the more complete implementation of the bug fix across affected files.", "best_issue_reason": null, - "best_pr_reason": "PR 47162 is the best overall representative since it covers the same concrete bug with a fuller implementation and validation, while 46310 is a narrower subset of the same change.", + "best_pr_reason": "PR 47162 is the strongest representative of this bug fix set because it contains the same concrete correction as 46310 plus broader coverage and validation, making it the most useful single artifact to point to.", "soft_edge_verdicts": [ { "left": "pull_request:46310", "right": "pull_request:47162", "accept": true, - "reason": "Same underlying off-by-one right-window boundary bug in MoonshineStreaming; the Moonshine diffs match and 47162 is a superset that could plausibly subsume 46310." + "reason": "Same underlying off-by-one right-window mask bug and same code path; 47162 includes the same Moonshine Streaming fix, so these are effectively overlapping fixes that could be merged into one PR." } ] }, @@ -28485,7 +28658,7 @@ "soft_pairs": [ "pull_request:46574|pull_request:46580" ], - "prepared_review_unit_hash": "700b0a3454d1c5020cc9762787d0e9fac56ab9344b8b3750d0a47e38bf774614", + "prepared_review_unit_hash": "72a4846d343e0e9ed90dcdca126d61b306fd0a1c68f721e6db7034705f38b72d", "component_budget": { "node_count": 2, "item_count": 2, @@ -28507,18 +28680,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate pull requests touching different tokenizer problems: CodeLlama decode stripping on leading spaces versus DeepSeek-Coder/DeepSeek-LLM loading the wrong tokenizer class. They should not be deduplicated.", - "confidence": 0.98, + "summary": "The two PRs are both tokenizer-related, but they fix different bugs in different code paths: CodeLlama decode stripping vs. DeepSeek model-class selection/loading. They should not be clustered as duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No single canonical PR for this cluster: the PRs address distinct tokenizer defects and are not mergeable into one change without conflating unrelated fixes.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "PR #46574 is the better representative of a concrete, self-contained bug fix with a clear failure mode and explicit issue link, while #46580 is a separate checkpoint-routing workaround.", "soft_edge_verdicts": [ { "left": "pull_request:46574", "right": "pull_request:46580", "accept": false, - "reason": "Different underlying bugs and code paths: 46574 changes CodeLlama tokenizer decode behavior, while 46580 changes AutoTokenizer checkpoint-prefix routing for DeepSeek checkpoints. They do not look mergeable into one PR." + "reason": "Different underlying problems and code paths: #46574 fixes CodeLlama decode behavior removing real leading spaces, while #46580 changes AutoTokenizer checkpoint-prefix routing for DeepSeek models loading the wrong tokenizer class." } ] }, @@ -28540,7 +28713,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-47122-10", + "cluster_id": "cluster-47122-13", "nodes": [ "pull_request:47122", "pull_request:47123", @@ -28551,7 +28724,10 @@ "pull_request:47281", "pull_request:47282", "pull_request:47310", - "pull_request:47380" + "pull_request:47380", + "pull_request:47443", + "pull_request:47444", + "pull_request:47445" ], "soft_pairs": [ "pull_request:47122|pull_request:47281", @@ -28560,198 +28736,306 @@ "pull_request:47122|pull_request:47224", "pull_request:47122|pull_request:47310", "pull_request:47123|pull_request:47280", + "pull_request:47123|pull_request:47443", "pull_request:47123|pull_request:47266", "pull_request:47123|pull_request:47281", - "pull_request:47123|pull_request:47124", "pull_request:47124|pull_request:47310", + "pull_request:47124|pull_request:47445", "pull_request:47124|pull_request:47282", "pull_request:47124|pull_request:47380", - "pull_request:47124|pull_request:47281", + "pull_request:47124|pull_request:47443", + "pull_request:47266|pull_request:47443", "pull_request:47266|pull_request:47280", - "pull_request:47266|pull_request:47380", - "pull_request:47266|pull_request:47281", - "pull_request:47124|pull_request:47266", + "pull_request:47266|pull_request:47444", + "pull_request:47266|pull_request:47445", + "pull_request:47280|pull_request:47443", "pull_request:47280|pull_request:47281", "pull_request:47280|pull_request:47310", - "pull_request:47280|pull_request:47380", - "pull_request:47281|pull_request:47282", + "pull_request:47281|pull_request:47443", + "pull_request:47281|pull_request:47445", "pull_request:47282|pull_request:47380", "pull_request:47282|pull_request:47310", "pull_request:47280|pull_request:47282", - "pull_request:47310|pull_request:47380" - ], - "prepared_review_unit_hash": "0373b990f385bdee34f58d47c5c1a83d3de6221a2eb2f266b13a195687a75d66", + "pull_request:47282|pull_request:47444", + "pull_request:47310|pull_request:47380", + "pull_request:47310|pull_request:47445", + "pull_request:47280|pull_request:47380", + "pull_request:47380|pull_request:47444", + "pull_request:47443|pull_request:47445", + "pull_request:47443|pull_request:47444", + "pull_request:47444|pull_request:47445", + "pull_request:47280|pull_request:47444", + "pull_request:47123|pull_request:47444", + "pull_request:47282|pull_request:47445" + ], + "prepared_review_unit_hash": "0e8e629c20e3ca76318adb85e414727fac470f4415fa350733a7ec43a9d1521c", "component_budget": { - "node_count": 10, - "item_count": 10, - "soft_pair_count": 25, - "serialized_chars": 28953, - "estimated_input_tokens": 7239, - "estimated_eval_tokens": 14734 + "node_count": 13, + "item_count": 13, + "soft_pair_count": 37, + "serialized_chars": 44721, + "estimated_input_tokens": 11181, + "estimated_eval_tokens": 22618 }, "budget": { - "node_count": 10, - "item_count": 10, - "soft_pair_count": 25, - "serialized_chars": 28953, - "estimated_input_tokens": 7239, - "estimated_eval_tokens": 14734 + "node_count": 13, + "item_count": 13, + "soft_pair_count": 37, + "serialized_chars": 44721, + "estimated_input_tokens": 11181, + "estimated_eval_tokens": 22618 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated, model-specific serge integration-test fix PRs. Only two soft-duplicate pairs look like true same-bug duplicates: the MiniMax-M3-VL load_error fixes and the Mamba2 OOM fixes.", - "confidence": 0.91, + "summary": "This cluster is mostly false-positive similarity across unrelated serge-generated test-fix PRs. True duplicates only appear within the MiniMax-M3-VL load_error pair, the Mamba2 OOM pair, and the DeepSeek rope/Yarn cleanup pair.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "47280 is the clearest representative of a true duplicate pair: it is the more complete MiniMax-M3-VL load_error fix and directly subsumes the earlier offload-folder workaround in the same test file.", + "canonical_pr_reason": "PR 47443 is the clearest representative of the shared DeepSeek/MoE rope/Yarn code cleanup that also overlaps 47444; it has the strongest common code-path change among the PRs.", "best_issue_reason": null, - "best_pr_reason": "47280 is the strongest standalone PR in this cluster because it targets one concrete failure mode cleanly and reads like the best canonical target among the true duplicates.", + "best_pr_reason": "PR 47443 is the best global representative because it carries the broad shared model-side fix across the DeepSeek/MoE files, while the nearby PRs are either narrower duplicates or mixed with unrelated changes.", "soft_edge_verdicts": [ { "left": "pull_request:47122", "right": "pull_request:47281", "accept": false, - "reason": "Different models and different changes; both are test/integration fixes, but not the same underlying bug or code path." + "reason": "Different models and unrelated fixes; only the automation/template looks similar." }, { "left": "pull_request:47122", "right": "pull_request:47123", "accept": false, - "reason": "Different models and failure modes: EdgeTAM import_or_config vs MiniMax-M3-VL load_error." + "reason": "Edgetam import_or_config bug vs MiniMax-M3-VL load_error; not the same change." }, { "left": "pull_request:47122", "right": "pull_request:47280", "accept": false, - "reason": "Different models; no shared concrete bug despite both being serge-generated test-fix PRs." + "reason": "Different model and failure; same test-fix pattern but not the same underlying bug." }, { "left": "pull_request:47122", "right": "pull_request:47224", "accept": false, - "reason": "47224 is a bundled multi-fix PR spanning unrelated files/models, so it is not the same underlying change as the focused EdgeTAM fix." + "reason": "47224 is a bundled multi-fix PR; the shared Edgetam hunk is not enough to treat them as the same PR." }, { "left": "pull_request:47122", "right": "pull_request:47310", "accept": false, - "reason": "Different models and code paths; same broad symptom class is not enough." + "reason": "Edgetam vs Gemma3; different code paths and failure modes." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47266", + "accept": false, + "reason": "MiniMax-M3-VL load_error vs DeepSeek-V3.2 load_error; same symptom, different model and fix." }, { "left": "pull_request:47123", "right": "pull_request:47280", "accept": true, - "reason": "Same MiniMax-M3-VL load_error in the same test file; 47280 is a fuller version of the offload-folder fix and could plausibly subsume 47123." + "reason": "Same MiniMax-M3-VL load_error; both add temporary offload-folder handling in the same loader/test path." }, { "left": "pull_request:47123", - "right": "pull_request:47266", + "right": "pull_request:47281", "accept": false, - "reason": "MiniMax-M3-VL load_error vs DeepSeek-V3.2 load_error are different models and likely different causes." + "reason": "MiniMax-M3-VL load_error vs glm46v/glm4v modeling change; not the same bug." }, { "left": "pull_request:47123", - "right": "pull_request:47281", + "right": "pull_request:47443", "accept": false, - "reason": "Different models and code paths; not the same bug." + "reason": "MiniMax test load_error and DeepSeek rope cleanup are unrelated fixes." }, { "left": "pull_request:47123", - "right": "pull_request:47124", + "right": "pull_request:47444", "accept": false, - "reason": "MiniMax-M3-VL load_error and Falcon-Mamba output mismatch are unrelated failures." + "reason": "Different models and different fixes; no shared concrete code-path problem." }, { "left": "pull_request:47124", "right": "pull_request:47310", "accept": false, - "reason": "Both are output-mismatch fixes, but for different models and different generation behavior/code paths." + "reason": "Falcon-Mamba output mismatch vs Gemma3 output mismatch; same symptom category only." + }, + { + "left": "pull_request:47124", + "right": "pull_request:47445", + "accept": false, + "reason": "Falcon-Mamba generation test vs PEFT loading fix; unrelated code paths." }, { "left": "pull_request:47124", "right": "pull_request:47282", "accept": false, - "reason": "Different models and failure modes; output mismatch is not the same as Mamba2 OOM." + "reason": "Falcon-Mamba output mismatch vs Mamba2 OOM; different underlying issues." }, { "left": "pull_request:47124", "right": "pull_request:47380", "accept": false, - "reason": "Different models and unrelated concrete fixes." + "reason": "Falcon-Mamba output mismatch vs Mamba2 OOM; only broad test-fix similarity." }, { "left": "pull_request:47124", - "right": "pull_request:47281", + "right": "pull_request:47443", + "accept": false, + "reason": "Falcon-Mamba generation output issue is unrelated to the DeepSeek rope cleanup." + }, + { + "left": "pull_request:47266", + "right": "pull_request:47443", "accept": false, - "reason": "Different models; same broad integration-test theme is too weak." + "reason": "DeepSeek-V3.2 load_error test workaround vs model implementation cleanup; not the same concrete bug." }, { "left": "pull_request:47266", "right": "pull_request:47280", "accept": false, - "reason": "Different models and different failure mechanisms; not the same bug." + "reason": "Different models and different load_error workarounds; no shared code-path." }, { "left": "pull_request:47266", - "right": "pull_request:47281", + "right": "pull_request:47444", "accept": false, - "reason": "DeepSeek-V3.2 load_error vs GLM46V model changes are unrelated." + "reason": "DeepSeek-V3.2 test load_error vs MPT model cleanup; unrelated." }, { "left": "pull_request:47266", - "right": "pull_request:47380", + "right": "pull_request:47445", "accept": false, - "reason": "Different models and different failure modes." + "reason": "DeepSeek-V3.2 load_error test fix vs PEFT integration fix; unrelated changes." }, { "left": "pull_request:47280", "right": "pull_request:47281", "accept": false, - "reason": "Same serge-generated style, but different models and unrelated fixes." + "reason": "MiniMax-M3-VL load_error vs glm46v/glm4v refactor; different models and bugs." }, { "left": "pull_request:47280", "right": "pull_request:47310", "accept": false, - "reason": "MiniMax-M3-VL load_error vs Gemma3 generation-config behavior are unrelated." + "reason": "MiniMax-M3-VL load_error vs Gemma3 output mismatch; not the same issue." }, { "left": "pull_request:47280", "right": "pull_request:47380", "accept": false, - "reason": "Different models and different concrete failures; not mergeable as one duplicate fix." + "reason": "Same broad OOM theme, but different model/test and different mitigation; not a duplicate." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47443", + "accept": false, + "reason": "MiniMax loader workaround vs DeepSeek rope cleanup; unrelated." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47444", + "accept": false, + "reason": "MiniMax loader workaround vs MPT/DeepSeek code cleanup; not the same bug." }, { "left": "pull_request:47281", - "right": "pull_request:47282", + "right": "pull_request:47443", + "accept": false, + "reason": "glm46v/glm4v model refactor vs DeepSeek rope cleanup; unrelated." + }, + { + "left": "pull_request:47281", + "right": "pull_request:47445", "accept": false, - "reason": "Different models and unrelated problems." + "reason": "glm46v/glm4v refactor vs PEFT integration change; different code paths." }, { "left": "pull_request:47282", "right": "pull_request:47380", "accept": true, - "reason": "Same Mamba2 OOM integration-test failure in the same file; both are memory-related fixes that could plausibly be merged into one PR." + "reason": "Same Mamba2 OOM in the same integration test; both adjust memory/runtime behavior to avoid the crash." }, { "left": "pull_request:47282", "right": "pull_request:47310", "accept": false, - "reason": "Mamba2 OOM vs Gemma3 output-mismatch are different concrete issues." + "reason": "Mamba2 OOM vs Gemma3 output mismatch; unrelated failures." }, { "left": "pull_request:47280", "right": "pull_request:47282", "accept": false, - "reason": "Different models and unrelated failures." + "reason": "MiniMax-M3-VL load_error and Mamba2 OOM are different model failures." + }, + { + "left": "pull_request:47282", + "right": "pull_request:47444", + "accept": false, + "reason": "Mamba2 OOM workaround vs MPT/DeepSeek cleanup; not the same change." + }, + { + "left": "pull_request:47310", + "right": "pull_request:47380", + "accept": false, + "reason": "Gemma3 output mismatch vs Mamba2 OOM; different underlying bugs." }, { "left": "pull_request:47310", + "right": "pull_request:47445", + "accept": false, + "reason": "Gemma3 generation/default-cache fix vs PEFT loading fix; unrelated." + }, + { + "left": "pull_request:47280", "right": "pull_request:47380", "accept": false, - "reason": "Different models and different fixes; only a broad test-failure theme is shared." + "reason": "MiniMax-M3-VL load_error and Mamba2 OOM are different model-specific issues." + }, + { + "left": "pull_request:47380", + "right": "pull_request:47444", + "accept": false, + "reason": "Mamba2 OOM mitigation vs DeepSeek/MPT code cleanup; unrelated." + }, + { + "left": "pull_request:47443", + "right": "pull_request:47445", + "accept": false, + "reason": "47445 adds a PEFT-specific load fix in addition to overlapping model-file edits; not the same concrete change." + }, + { + "left": "pull_request:47443", + "right": "pull_request:47444", + "accept": true, + "reason": "They share the same DeepSeek/MoE rope/Yarn cleanup across the same model files; only the targeted failing test differs." + }, + { + "left": "pull_request:47444", + "right": "pull_request:47445", + "accept": false, + "reason": "Shared model-file overlap, but 47445 is mixed with an unrelated PEFT integration fix." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47444", + "accept": false, + "reason": "MiniMax-M3-VL loader workaround vs DeepSeek/MPT code cleanup; not the same bug." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47444", + "accept": false, + "reason": "MiniMax-M3-VL load_error workaround vs MPT/DeepSeek cleanup; different underlying issue." + }, + { + "left": "pull_request:47282", + "right": "pull_request:47445", + "accept": false, + "reason": "Mamba2 OOM workaround vs PEFT integration and model-file cleanup; not the same change." } ] }, @@ -28781,7 +29065,7 @@ "soft_pairs": [ "pull_request:47140|pull_request:47303" ], - "prepared_review_unit_hash": "c6e7017a4cdd2c00cf0cc34e9f6f46ca95996fec6842ff488f57c8fc297c306d", + "prepared_review_unit_hash": "189292c1154dd9dfd6e9871e379a602e14017e8fb9bb897b5cefada8431c7e32", "component_budget": { "node_count": 2, "item_count": 2, @@ -28803,24 +29087,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs appear to fix the same chat CLI base-URL bug around management endpoints when the server URL includes /v1, with one normalizing the service root centrally and the other routing load_model to the service root.", + "summary": "Both PRs address the same chat CLI bug where management endpoints are built from a `/v1` base URL instead of the service root. PR #47140 is the broader fix; PR #47303 is a narrower variant for the same path.", "confidence": 0.82, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 47140 is the cleaner canonical fix: it normalizes the chat client base URL once and makes all management calls use the service root, directly addressing the /v1 handling bug.", + "canonical_pr_reason": "PR #47140 is the better canonical representative because it fixes the base URL normalization once in `RichInterface`, covering management calls consistently rather than patching only one call site.", "best_issue_reason": null, - "best_pr_reason": "PR 47140 is the best overall match because it fixes the underlying URL normalization in one place instead of patching a single call site.", + "best_pr_reason": "PR #47140 is the strongest overall candidate: it directly addresses the `/v1` base handling at initialization and appears more complete for the underlying routing bug.", "soft_edge_verdicts": [ { "left": "pull_request:47140", "right": "pull_request:47303", "accept": true, - "reason": "Both PRs target the same concrete chat management URL problem for /v1 service URLs and are plausibly mergeable as alternative implementations of the same fix." + "reason": "Same underlying bug: chat management URLs are incorrectly rooted at `/v1`. Both PRs fix that service-root routing in `src/transformers/cli/chat.py`, and they could be consolidated into one change by choosing one implementation approach." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is supported by the packet: both PRs address the chat CLI\u2019s `/v1` base-URL handling for management endpoints, and the soft-edge accept is conservative because the changes are clearly in the same file, overlap the same bug area, and look mergeable into one fix. The wording stays within the evidence shown." + "feedback": "The summary is grounded in the packet: both PRs target chat CLI management URL routing away from a `/v1` base, and the distinction between a broader centralized fix (#47140) and a narrower call-site fix (#47303) is supported by the diffs and titles. The soft-edge mergeability verdict is conservative enough: they appear to address the same bug and could be combined into one PR by selecting one implementation approach rather than both." }, "evaluator_used": true, "retried": false, @@ -28847,12 +29131,12 @@ "soft_pairs": [ "pull_request:47261|pull_request:47262" ], - "prepared_review_unit_hash": "d64d8746ce05c60bcbd294badd1357a23385bf4948329f6e4b2fec075f979cf6", + "prepared_review_unit_hash": "7a46b43c3f55ce21663956bacbe3ba98b57dfe48acd1bfb1d0d3dd0839c7276f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4418, + "serialized_chars": 4419, "estimated_input_tokens": 1105, "estimated_eval_tokens": 2466 }, @@ -28860,7 +29144,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4418, + "serialized_chars": 4419, "estimated_input_tokens": 1105, "estimated_eval_tokens": 2466 }, @@ -28869,18 +29153,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two pull requests are unrelated: one fixes loss scaling with label smoothing during gradient accumulation, while the other adds/enforces min_tokens_to_keep in TopHLogitsWarper. They do not appear to be duplicate fixes for the same underlying bug.", + "summary": "The cluster contains two unrelated open PRs: one fixes gradient inflation with label smoothing during gradient accumulation, and the other adds/validates `min_tokens_to_keep` handling in `TopHLogitsWarper`. They target different code paths and are not duplicates.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 47261 is the stronger canonical PR for the cluster because it has an explicit issue link (#46731) and a narrowly defined training-loss bug fix with clear code-path changes.", + "canonical_pr_reason": "PR 47261 is the stronger canonical representative because it addresses a concrete training bug, has an explicit issue link, and its diff spans the actual fix plus tests; the other PR is a separate generation feature/fix.", "best_issue_reason": null, - "best_pr_reason": "PR 47261 is the best representative PR overall because it is the more specific, well-scoped bug fix and is directly tied to an issue; PR 47262 addresses a separate generation-time validation behavior.", + "best_pr_reason": "PR 47261 is the best overall representative of the cluster since it is more clearly bug-driven and self-contained, while PR 47262 is unrelated in subsystem and behavior.", "soft_edge_verdicts": [ { "left": "pull_request:47261", "right": "pull_request:47262", "accept": false, - "reason": "Reject: these touch different subsystems and fix different problems (trainer label smoothing/gradient accumulation vs. TopHLogitsWarper parameter validation). They are not the same underlying bug or mergeable into one PR." + "reason": "Different subsystems and behaviors: one changes Trainer/LabelSmoother loss scaling for gradient accumulation, the other changes generation logits warping to respect `min_tokens_to_keep`. They do not fix the same underlying bug and would not plausibly merge into one PR." } ] }, @@ -28910,12 +29194,12 @@ "soft_pairs": [ "pull_request:47315|pull_request:47430" ], - "prepared_review_unit_hash": "6c59c5933435d58ac02e0eab252dd6b0ea0c853b790d82e7d3d5003ba6bf8cb2", + "prepared_review_unit_hash": "8858cd1794ea2e0b7a8ea369a7e165f385b808df7437418d147d8559f3da0795", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4382, + "serialized_chars": 4383, "estimated_input_tokens": 1096, "estimated_eval_tokens": 2448 }, @@ -28923,7 +29207,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4382, + "serialized_chars": 4383, "estimated_input_tokens": 1096, "estimated_eval_tokens": 2448 }, @@ -28932,18 +29216,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs touch `compressed_tensors` MoE decompression, but they fix different bugs: one repairs shape reconstruction for non-32-divisible bit widths, while the other propagates `zero_point` for asymmetric quantization. Related area, not a duplicate change.", - "confidence": 0.96, + "summary": "Both PRs touch compressed_tensors MoE expert decompression, but they address different bugs: one fixes fallback shape reconstruction for non-32-divisible bit widths, the other preserves asymmetric zero_point handling. Related subsystem, not the same change.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "`pull_request:47315` is the cleaner representative of the cluster topic: it fixes a concrete decompression bug in the low-bit shape fallback, with a narrowly scoped change and a clear failure mode.", + "canonical_pr_reason": "PR 47430 is the broader decompression fix: it updates both the integration path and the quantizer to carry zero_point through asymmetric MoE expert decompression, whereas PR 47315 only patches shape inference for certain bit widths.", "best_issue_reason": null, - "best_pr_reason": "`pull_request:47315` is the strongest standalone PR because it addresses a specific, reproducible decompression error and is more directly tied to the underlying code-path bug than the zero-point propagation fix.", + "best_pr_reason": "PR 47430 is the better representative of the cluster because it covers a more complete decompression-path change across multiple files, while PR 47315 is a narrower single-file bug fix.", "soft_edge_verdicts": [ { "left": "pull_request:47315", "right": "pull_request:47430", "accept": false, - "reason": "Different underlying bugs in the same subsystem: 47315 fixes incorrect `weight_shape` fallback for 3/5/6/7-bit packing, while 47430 fixes missing `zero_point` handling for asymmetric MoE expert decompression. Not the same change and not safely mergeable into one PR." + "reason": "Different concrete defects: 47315 fixes incorrect unpacked-dimension fallback for 3/5/6/7-bit packing, while 47430 fixes missing zero_point propagation for asymmetric decompression. Same area, not the same underlying bug." } ] }, @@ -28973,12 +29257,12 @@ "soft_pairs": [ "pull_request:47400|pull_request:47421" ], - "prepared_review_unit_hash": "2bf09c643e95f3b850f48e015ae5a7af1b0d6b29cff8d9ffb170fc752d19bb27", + "prepared_review_unit_hash": "ea1b59c27dbb00398b91a2f10f7e4e2137e4fe5d17f26d75cdaf52a2d433d604", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4618, + "serialized_chars": 4619, "estimated_input_tokens": 1155, "estimated_eval_tokens": 2566 }, @@ -28986,7 +29270,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4618, + "serialized_chars": 4619, "estimated_input_tokens": 1155, "estimated_eval_tokens": 2566 }, @@ -28995,18 +29279,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate GGUF-loading feature PRs for different model families (Phi-2 vs DeepSeek-V3). They overlap in the same integration/docs/tests files, but the underlying changes are not the same bug or change.", - "confidence": 0.98, + "summary": "Two open PRs touch the same GGUF loading/documentation/test area, but they add support for different model families: Phi-2 vs DeepSeek-V3. They are related by subsystem, not by the same underlying change.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: #47400 adds Phi-2 GGUF support, while #47421 adds DeepSeek-V3 support; they share infrastructure but target different architectures.", + "canonical_pr_reason": "PR #47421 is the broader representative of the cluster: it adds GGUF support for DeepSeek-V3 with a larger code path update and docs/test changes, making it a better anchor than the narrower Phi-2-only PR.", "best_issue_reason": null, - "best_pr_reason": "Neither PR is a duplicate representative of the other; both are valid but distinct feature additions, so there is no global best duplicate PR.", + "best_pr_reason": "PR #47421 is the best PR candidate because it appears to be the more substantial and more representative GGUF integration change, while #47400 is a separate model-specific addition.", "soft_edge_verdicts": [ { "left": "pull_request:47400", "right": "pull_request:47421", "accept": false, - "reason": "Different GGUF model-support additions for different architectures; shared files reflect common loader plumbing, not the same change." + "reason": "Both PRs modify the same GGUF integration files, but they target different model architectures (Phi-2 vs DeepSeek-V3) and are not the same concrete bug or change." } ] }, @@ -29036,12 +29320,12 @@ "soft_pairs": [ "pull_request:47431|pull_request:47434" ], - "prepared_review_unit_hash": "024f1e398fde87cb34506572df1ea3f4edf83b8fe65905a2392e2caf201c62c5", + "prepared_review_unit_hash": "13cba77cceee4848611f83b8a4724d5c5199e766bfc5c2744c9be77c7bb95037", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3810, + "serialized_chars": 3811, "estimated_input_tokens": 953, "estimated_eval_tokens": 2162 }, @@ -29049,7 +29333,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3810, + "serialized_chars": 3811, "estimated_input_tokens": 953, "estimated_eval_tokens": 2162 }, @@ -29058,18 +29342,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests are the same dependency-bound relaxation for tokenizers, with identical diffs and the same target issue (#47429). PR 47434 is the better canonical PR because its title is clearer and it explicitly closes the issue.", + "summary": "Both pull requests make the same tokenizers upper-bound relaxation in the same two files and target the same issue, so they are duplicates of one another.", "confidence": 0.99, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:47434 is the clearest canonical PR: it has the most explicit title, directly references the tracked issue, and the diff matches the other PR exactly.", + "canonical_pr_reason": "PR #47434 is the better canonical PR because its title is more explicit and it directly references the tracked issue (#47429), while the diff is identical to #47431.", "best_issue_reason": null, - "best_pr_reason": "pull_request:47434 is the best PR candidate because it is explicit about fixing #47429 and is identical in scope and code changes to pull_request:47431.", + "best_pr_reason": "PR #47434 is the strongest representative of the cluster: it states the exact change, links the issue, and has the same code edits as #47431.", "soft_edge_verdicts": [ { "left": "pull_request:47431", "right": "pull_request:47434", "accept": true, - "reason": "Accept: these are duplicate PRs with identical file edits and the same underlying change\u2014relaxing the tokenizers upper bound from <=0.23.0 to <0.24.0 to allow 0.23.1." + "reason": "Identical diff, same touched files, and the same explicit issue target; these are the same underlying change and could be merged as one PR." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index d2b09ff6a8aa8be32cac7dc4246e3e6822bb27d4..878148a6db8bfa8e96c99aa4686c1edf6e84edd1 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260721t000056z", + "analysis_id": "hybrid-model-20260721t060018z", "archived_artifacts": { - "hybrid": "snapshots/20260721T000056Z/analysis-runs/hybrid-model-20260721t000056z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260721T000056Z/analysis-runs/hybrid-model-20260721t000056z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-07-21T00:24:04Z", + "published_at": "2026-07-21T06:15:11Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260721T000056Z", + "snapshot_id": "20260721T060018Z", "variant": "hybrid" } diff --git a/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.json b/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..bc109b7807859d41ab841a38ece715741e6df3fb --- /dev/null +++ b/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.json @@ -0,0 +1,3772 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260721T060018Z", + "generated_at": "2026-07-21T06:15:09Z", + "evidence_quality": "full", + "llm_enrichment": true, + "meta_bugs": [ + { + "cluster_id": "cluster-43979-11", + "summary": "Cluster of 1 issues and 10 PRs centered on issue #43979.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 43979, + "canonical_pr_number": 44007, + "issue_numbers": [ + 43979 + ], + "pr_numbers": [ + 43996, + 44007, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43996, + "right_pr_number": 44007, + "code_similarity": 0.179, + "size_similarity": 0.576, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.429, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44013, + "code_similarity": 0.122, + "size_similarity": 0.318, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.392, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44044, + "code_similarity": 0.245, + "size_similarity": 0.864, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.479, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44066, + "code_similarity": 0.225, + "size_similarity": 0.818, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.408, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44072, + "code_similarity": 0.14, + "size_similarity": 0.303, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.528, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44085, + "code_similarity": 0.216, + "size_similarity": 0.783, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.398, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44129, + "code_similarity": 0.163, + "size_similarity": 0.643, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.229, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44154, + "code_similarity": 0.153, + "size_similarity": 0.535, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.31, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44722, + "code_similarity": 0.225, + "size_similarity": 0.848, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.368, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44013, + "code_similarity": 0.19, + "size_similarity": 0.553, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.531, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44044, + "code_similarity": 0.186, + "size_similarity": 0.667, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.354, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44066, + "code_similarity": 0.188, + "size_similarity": 0.704, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.315, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44072, + "code_similarity": 0.212, + "size_similarity": 0.526, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.708, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44085, + "code_similarity": 0.195, + "size_similarity": 0.735, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.318, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44129, + "code_similarity": 0.103, + "size_similarity": 0.37, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.191, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44154, + "code_similarity": 0.238, + "size_similarity": 0.93, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.344, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44722, + "code_similarity": 0.178, + "size_similarity": 0.679, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.28, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44044, + "code_similarity": 0.126, + "size_similarity": 0.368, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.351, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44066, + "code_similarity": 0.127, + "size_similarity": 0.389, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.325, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44072, + "code_similarity": 0.29, + "size_similarity": 0.952, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.667, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44085, + "code_similarity": 0.131, + "size_similarity": 0.406, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.329, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44129, + "code_similarity": 0.07, + "size_similarity": 0.205, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.192, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44154, + "code_similarity": 0.177, + "size_similarity": 0.594, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.389, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44722, + "code_similarity": 0.118, + "size_similarity": 0.375, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.287, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44066, + "code_similarity": 0.25, + "size_similarity": 0.947, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.404, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44072, + "code_similarity": 0.136, + "size_similarity": 0.351, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.442, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44085, + "code_similarity": 0.24, + "size_similarity": 0.906, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.394, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44129, + "code_similarity": 0.147, + "size_similarity": 0.555, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.243, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44154, + "code_similarity": 0.17, + "size_similarity": 0.62, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.306, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44722, + "code_similarity": 0.257, + "size_similarity": 0.982, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.402, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44072, + "code_similarity": 0.133, + "size_similarity": 0.37, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.393, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44085, + "code_similarity": 0.763, + "size_similarity": 0.957, + "file_overlap": 0.5, + "area_overlap": 0.825, + "patch_similarity": 0.887, + "shared_filenames": [ + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 41 + ], + [ + 174, + 181 + ], + [ + 250, + 257 + ], + [ + 398, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 466 + ], + [ + 471, + 488 + ], + [ + 518, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 606 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ], + "right_ranges": [ + [ + 33, + 40 + ], + [ + 173, + 180 + ], + [ + 249, + 256 + ], + [ + 397, + 405 + ], + [ + 408, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 465 + ], + [ + 470, + 489 + ], + [ + 517, + 539 + ], + [ + 553, + 559 + ], + [ + 565, + 574 + ], + [ + 579, + 597 + ], + [ + 600, + 611 + ], + [ + 633, + 639 + ], + [ + 645, + 652 + ], + [ + 657, + 673 + ], + [ + 716, + 728 + ], + [ + 737, + 743 + ], + [ + 748, + 771 + ], + [ + 789, + 794 + ] + ] + } + ] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44129, + "code_similarity": 0.145, + "size_similarity": 0.526, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.263, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44154, + "code_similarity": 0.174, + "size_similarity": 0.654, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.286, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44722, + "code_similarity": 0.907, + "size_similarity": 0.964, + "file_overlap": 1.0, + "area_overlap": 0.808, + "patch_similarity": 0.874, + "shared_filenames": [ + "src/transformers/models/codegen/modeling_codegen.py", + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/codegen/modeling_codegen.py", + "left_ranges": [ + [ + 245, + 266 + ] + ], + "right_ranges": [ + [ + 228, + 234 + ] + ] + }, + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 41 + ], + [ + 174, + 181 + ], + [ + 250, + 257 + ], + [ + 398, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 466 + ], + [ + 471, + 488 + ], + [ + 518, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 606 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ], + "right_ranges": [ + [ + 33, + 42 + ], + [ + 175, + 182 + ], + [ + 251, + 258 + ], + [ + 399, + 420 + ], + [ + 426, + 435 + ], + [ + 460, + 467 + ], + [ + 472, + 489 + ], + [ + 519, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 612 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 663 + ], + [ + 666, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ] + } + ] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44085, + "code_similarity": 0.137, + "size_similarity": 0.387, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.398, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44129, + "code_similarity": 0.074, + "size_similarity": 0.195, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.231, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44154, + "code_similarity": 0.175, + "size_similarity": 0.566, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.414, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44722, + "code_similarity": 0.124, + "size_similarity": 0.357, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.347, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44129, + "code_similarity": 0.141, + "size_similarity": 0.503, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.272, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44154, + "code_similarity": 0.18, + "size_similarity": 0.684, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.289, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44722, + "code_similarity": 0.728, + "size_similarity": 0.923, + "file_overlap": 0.5, + "area_overlap": 0.791, + "patch_similarity": 0.78, + "shared_filenames": [ + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 40 + ], + [ + 173, + 180 + ], + [ + 249, + 256 + ], + [ + 397, + 405 + ], + [ + 408, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 465 + ], + [ + 470, + 489 + ], + [ + 517, + 539 + ], + [ + 553, + 559 + ], + [ + 565, + 574 + ], + [ + 579, + 597 + ], + [ + 600, + 611 + ], + [ + 633, + 639 + ], + [ + 645, + 652 + ], + [ + 657, + 673 + ], + [ + 716, + 728 + ], + [ + 737, + 743 + ], + [ + 748, + 771 + ], + [ + 789, + 794 + ] + ], + "right_ranges": [ + [ + 33, + 42 + ], + [ + 175, + 182 + ], + [ + 251, + 258 + ], + [ + 399, + 420 + ], + [ + 426, + 435 + ], + [ + 460, + 467 + ], + [ + 472, + 489 + ], + [ + 519, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 612 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 663 + ], + [ + 666, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ] + } + ] + }, + { + "left_pr_number": 44129, + "right_pr_number": 44154, + "code_similarity": 0.099, + "size_similarity": 0.344, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.199, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44129, + "right_pr_number": 44722, + "code_similarity": 0.146, + "size_similarity": 0.545, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.247, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44154, + "right_pr_number": 44722, + "code_similarity": 0.164, + "size_similarity": 0.631, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.255, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-46894-7", + "summary": "Cluster of 3 issues and 4 PRs centered on issue #46894.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 46894, + "canonical_pr_number": 46903, + "issue_numbers": [ + 46894, + 46897, + 46901 + ], + "pr_numbers": [ + 46895, + 46903, + 46958, + 47090 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46895, + "right_pr_number": 46903, + "code_similarity": 0.533, + "size_similarity": 0.404, + "file_overlap": 0.333, + "area_overlap": 0.933, + "patch_similarity": 0.17, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 637, + 650 + ] + ], + "right_ranges": [ + [ + 637, + 651 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 500, + 513 + ] + ], + "right_ranges": [ + [ + 500, + 514 + ] + ] + } + ] + }, + { + "left_pr_number": 46895, + "right_pr_number": 46958, + "code_similarity": 0.779, + "size_similarity": 0.915, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.81, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "tests/models/cohere_asr/test_modeling_cohere_asr.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 637, + 650 + ] + ], + "right_ranges": [ + [ + 22, + 28 + ], + [ + 638, + 645 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 500, + 513 + ] + ], + "right_ranges": [ + [ + 16, + 22 + ], + [ + 501, + 508 + ] + ] + }, + { + "filename": "tests/models/cohere_asr/test_modeling_cohere_asr.py", + "left_ranges": [ + [ + 35, + 41 + ], + [ + 176, + 214 + ] + ], + "right_ranges": [ + [ + 15, + 21 + ], + [ + 180, + 221 + ] + ] + } + ] + }, + { + "left_pr_number": 46895, + "right_pr_number": 47090, + "code_similarity": 0.459, + "size_similarity": 0.281, + "file_overlap": 0.273, + "area_overlap": 0.765, + "patch_similarity": 0.355, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "tests/models/cohere_asr/test_modeling_cohere_asr.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 637, + 650 + ] + ], + "right_ranges": [ + [ + 637, + 649 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 500, + 513 + ] + ], + "right_ranges": [ + [ + 500, + 512 + ] + ] + }, + { + "filename": "tests/models/cohere_asr/test_modeling_cohere_asr.py", + "left_ranges": [ + [ + 35, + 41 + ], + [ + 176, + 214 + ] + ], + "right_ranges": [ + [ + 175, + 224 + ] + ] + } + ] + }, + { + "left_pr_number": 46903, + "right_pr_number": 46958, + "code_similarity": 0.325, + "size_similarity": 0.369, + "file_overlap": 0.333, + "area_overlap": 0.364, + "patch_similarity": 0.156, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 637, + 651 + ] + ], + "right_ranges": [ + [ + 22, + 28 + ], + [ + 638, + 645 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 500, + 514 + ] + ], + "right_ranges": [ + [ + 16, + 22 + ], + [ + 501, + 508 + ] + ] + } + ] + }, + { + "left_pr_number": 46903, + "right_pr_number": 47090, + "code_similarity": 0.576, + "size_similarity": 0.695, + "file_overlap": 0.333, + "area_overlap": 0.871, + "patch_similarity": 0.214, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "src/transformers/models/pp_formulanet/modeling_pp_formulanet.py", + "src/transformers/models/pp_formulanet/modular_pp_formulanet.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 637, + 651 + ] + ], + "right_ranges": [ + [ + 637, + 649 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 500, + 514 + ] + ], + "right_ranges": [ + [ + 500, + 512 + ] + ] + }, + { + "filename": "src/transformers/models/pp_formulanet/modeling_pp_formulanet.py", + "left_ranges": [ + [ + 1103, + 1116 + ] + ], + "right_ranges": [ + [ + 1103, + 1118 + ] + ] + }, + { + "filename": "src/transformers/models/pp_formulanet/modular_pp_formulanet.py", + "left_ranges": [ + [ + 488, + 501 + ] + ], + "right_ranges": [ + [ + 488, + 503 + ] + ] + } + ] + }, + { + "left_pr_number": 46958, + "right_pr_number": 47090, + "code_similarity": 0.397, + "size_similarity": 0.257, + "file_overlap": 0.273, + "area_overlap": 0.598, + "patch_similarity": 0.362, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "tests/models/cohere_asr/test_modeling_cohere_asr.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 22, + 28 + ], + [ + 638, + 645 + ] + ], + "right_ranges": [ + [ + 637, + 649 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 16, + 22 + ], + [ + 501, + 508 + ] + ], + "right_ranges": [ + [ + 500, + 512 + ] + ] + }, + { + "filename": "tests/models/cohere_asr/test_modeling_cohere_asr.py", + "left_ranges": [ + [ + 15, + 21 + ], + [ + 180, + 221 + ] + ], + "right_ranges": [ + [ + 175, + 224 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43451-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #47321.", + "status": "open", + "confidence": 0.6, + "canonical_issue_number": 47321, + "canonical_pr_number": 47322, + "issue_numbers": [ + 47321 + ], + "pr_numbers": [ + 43451, + 47322 + ], + "evidence_types": [ + "closing_reference", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43451, + "right_pr_number": 47322, + "code_similarity": 0.001, + "size_similarity": 0.002, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.003, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-41211-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41211.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 41211, + "canonical_pr_number": 44339, + "issue_numbers": [ + 41211 + ], + "pr_numbers": [ + 41356, + 44339 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41356, + "right_pr_number": 44339, + "code_similarity": 0.155, + "size_similarity": 0.094, + "file_overlap": 0.273, + "area_overlap": 0.078, + "patch_similarity": 0.183, + "shared_filenames": [ + "docs/source/en/model_doc/deimv2.md", + "src/transformers/models/deimv2/__init__.py", + "src/transformers/models/deimv2/configuration_deimv2.py", + "src/transformers/models/deimv2/modeling_deimv2.py", + "tests/models/deimv2/__init__.py", + "tests/models/deimv2/test_modeling_deimv2.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/model_doc/deimv2.md", + "left_ranges": [ + [ + 1, + 132 + ] + ], + "right_ranges": [ + [ + 1, + 65 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/__init__.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 29 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/configuration_deimv2.py", + "left_ranges": [ + [ + 1, + 74 + ] + ], + "right_ranges": [ + [ + 1, + 266 + ] + ] + }, + { + "filename": "tests/models/deimv2/test_modeling_deimv2.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 1734 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46509-5", + "summary": "Cluster of 5 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47143, + "issue_numbers": [], + "pr_numbers": [ + 46509, + 47143, + 47163, + 47301, + 47387 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46509, + "right_pr_number": 47143, + "code_similarity": 0.254, + "size_similarity": 0.793, + "file_overlap": 0.071, + "area_overlap": 0.143, + "patch_similarity": 0.161, + "shared_filenames": [ + "src/transformers/models/auto/auto_mappings.py", + "src/transformers/models/auto/modeling_auto.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/auto/auto_mappings.py", + "left_ranges": [ + [ + 412, + 418 + ], + [ + 1077, + 1083 + ], + [ + 1206, + 1212 + ] + ], + "right_ranges": [ + [ + 412, + 418 + ], + [ + 975, + 981 + ] + ] + }, + { + "filename": "src/transformers/models/auto/modeling_auto.py", + "left_ranges": [ + [ + 1097, + 1103 + ] + ], + "right_ranges": [ + [ + 348, + 354 + ], + [ + 1960, + 1966 + ] + ] + } + ] + }, + { + "left_pr_number": 46509, + "right_pr_number": 47163, + "code_similarity": 0.006, + "size_similarity": 0.012, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.024, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 46509, + "right_pr_number": 47301, + "code_similarity": 0.01, + "size_similarity": 0.013, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.047, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 46509, + "right_pr_number": 47387, + "code_similarity": 0.038, + "size_similarity": 0.119, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.095, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47143, + "right_pr_number": 47163, + "code_similarity": 0.004, + "size_similarity": 0.009, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.013, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47143, + "right_pr_number": 47301, + "code_similarity": 0.006, + "size_similarity": 0.01, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.026, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47143, + "right_pr_number": 47387, + "code_similarity": 0.028, + "size_similarity": 0.095, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.06, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47163, + "right_pr_number": 47301, + "code_similarity": 0.192, + "size_similarity": 0.933, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.033, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47163, + "right_pr_number": 47387, + "code_similarity": 0.025, + "size_similarity": 0.098, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.033, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47301, + "right_pr_number": 47387, + "code_similarity": 0.03, + "size_similarity": 0.105, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.059, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-47332-4", + "summary": "Cluster of 2 issues and 2 PRs centered on issue #47333.", + "status": "open", + "confidence": 0.6, + "canonical_issue_number": 47333, + "canonical_pr_number": 47336, + "issue_numbers": [ + 47332, + 47333 + ], + "pr_numbers": [ + 47336, + 47383 + ], + "evidence_types": [ + "closing_reference", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47336, + "right_pr_number": 47383, + "code_similarity": 0.019, + "size_similarity": 0.091, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.005, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-47138-4", + "summary": "Cluster of 2 issues and 2 PRs centered on issue #47138.", + "status": "open", + "confidence": 0.6, + "canonical_issue_number": 47138, + "canonical_pr_number": 47303, + "issue_numbers": [ + 47138, + 47139 + ], + "pr_numbers": [ + 47140, + 47303 + ], + "evidence_types": [ + "closing_reference", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47140, + "right_pr_number": 47303, + "code_similarity": 0.656, + "size_similarity": 0.65, + "file_overlap": 1.0, + "area_overlap": 0.447, + "patch_similarity": 0.466, + "shared_filenames": [ + "src/transformers/cli/chat.py", + "tests/cli/test_chat.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/chat.py", + "left_ranges": [ + [ + 20, + 26 + ], + [ + 103, + 122 + ], + [ + 346, + 356 + ], + [ + 383, + 389 + ], + [ + 473, + 479 + ] + ], + "right_ranges": [ + [ + 103, + 115 + ], + [ + 216, + 224 + ], + [ + 383, + 389 + ] + ] + }, + { + "filename": "tests/cli/test_chat.py", + "left_ranges": [ + [ + 15, + 28 + ], + [ + 51, + 105 + ] + ], + "right_ranges": [ + [ + 14, + 29 + ], + [ + 52, + 76 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43240, + "canonical_pr_number": 43251, + "issue_numbers": [ + 43240 + ], + "pr_numbers": [ + 43251, + 43254 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43251, + "right_pr_number": 43254, + "code_similarity": 0.794, + "size_similarity": 0.64, + "file_overlap": 1.0, + "area_overlap": 0.667, + "patch_similarity": 0.882, + "shared_filenames": [ + "src/transformers/loss/loss_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/loss/loss_utils.py", + "left_ranges": [ + [ + 30, + 50 + ] + ], + "right_ranges": [ + [ + 30, + 43 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45647-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45647.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 45647, + "canonical_pr_number": 45738, + "issue_numbers": [ + 45647 + ], + "pr_numbers": [ + 45737, + 45738 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 45737, + "right_pr_number": 45738, + "code_similarity": 0.526, + "size_similarity": 0.11, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.024, + "shared_filenames": [ + "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py", + "left_ranges": [ + [ + 491, + 497 + ] + ], + "right_ranges": [ + [ + 491, + 497 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46159-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46159.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 46159, + "canonical_pr_number": 46174, + "issue_numbers": [ + 46159 + ], + "pr_numbers": [ + 46174, + 46780 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 46174, + "right_pr_number": 46780, + "code_similarity": 0.841, + "size_similarity": 0.87, + "file_overlap": 1.0, + "area_overlap": 0.767, + "patch_similarity": 0.657, + "shared_filenames": [ + "src/transformers/models/aimv2/modeling_aimv2.py", + "src/transformers/models/d_fine/modeling_d_fine.py", + "src/transformers/models/deimv2/modeling_deimv2.py", + "src/transformers/models/pp_doclayout_v2/modeling_pp_doclayout_v2.py", + "src/transformers/models/pp_doclayout_v3/modeling_pp_doclayout_v3.py", + "src/transformers/models/rt_detr/modeling_rt_detr.py", + "src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py", + "src/transformers/models/vit_mae/modeling_vit_mae.py", + "src/transformers/models/vit_mae/modular_vit_mae.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/aimv2/modeling_aimv2.py", + "left_ranges": [ + [ + 134, + 141 + ], + [ + 144, + 154 + ], + [ + 157, + 165 + ] + ], + "right_ranges": [ + [ + 134, + 140 + ], + [ + 143, + 153 + ], + [ + 156, + 162 + ] + ] + }, + { + "filename": "src/transformers/models/d_fine/modeling_d_fine.py", + "left_ranges": [ + [ + 631, + 638 + ], + [ + 641, + 651 + ], + [ + 654, + 662 + ] + ], + "right_ranges": [ + [ + 631, + 637 + ], + [ + 640, + 650 + ], + [ + 653, + 659 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/modeling_deimv2.py", + "left_ranges": [ + [ + 703, + 710 + ], + [ + 713, + 723 + ], + [ + 726, + 734 + ] + ], + "right_ranges": [ + [ + 703, + 709 + ], + [ + 712, + 722 + ], + [ + 725, + 731 + ] + ] + }, + { + "filename": "src/transformers/models/pp_doclayout_v2/modeling_pp_doclayout_v2.py", + "left_ranges": [ + [ + 1551, + 1558 + ], + [ + 1561, + 1571 + ], + [ + 1574, + 1582 + ] + ], + "right_ranges": [ + [ + 1551, + 1557 + ], + [ + 1560, + 1570 + ], + [ + 1573, + 1579 + ] + ] + }, + { + "filename": "src/transformers/models/pp_doclayout_v3/modeling_pp_doclayout_v3.py", + "left_ranges": [ + [ + 804, + 811 + ], + [ + 814, + 824 + ], + [ + 827, + 835 + ] + ], + "right_ranges": [ + [ + 804, + 810 + ], + [ + 813, + 823 + ], + [ + 826, + 832 + ] + ] + }, + { + "filename": "src/transformers/models/rt_detr/modeling_rt_detr.py", + "left_ranges": [ + [ + 847, + 854 + ], + [ + 857, + 867 + ], + [ + 870, + 878 + ] + ], + "right_ranges": [ + [ + 847, + 853 + ], + [ + 856, + 866 + ], + [ + 869, + 875 + ] + ] + }, + { + "filename": "src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py", + "left_ranges": [ + [ + 976, + 983 + ], + [ + 986, + 996 + ], + [ + 999, + 1007 + ] + ], + "right_ranges": [ + [ + 976, + 982 + ], + [ + 985, + 995 + ], + [ + 998, + 1004 + ] + ] + }, + { + "filename": "src/transformers/models/vit_mae/modeling_vit_mae.py", + "left_ranges": [ + [ + 152, + 159 + ], + [ + 162, + 172 + ], + [ + 175, + 183 + ] + ], + "right_ranges": [ + [ + 152, + 158 + ], + [ + 161, + 171 + ], + [ + 174, + 180 + ] + ] + }, + { + "filename": "src/transformers/models/vit_mae/modular_vit_mae.py", + "left_ranges": [ + [ + 62, + 69 + ], + [ + 72, + 82 + ], + [ + 85, + 93 + ] + ], + "right_ranges": [ + [ + 62, + 68 + ], + [ + 71, + 81 + ], + [ + 84, + 90 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46869-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46869.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 46869, + "canonical_pr_number": 46886, + "issue_numbers": [ + 46869 + ], + "pr_numbers": [ + 46886, + 46922 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 46886, + "right_pr_number": 46922, + "code_similarity": 0.711, + "size_similarity": 0.857, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.429, + "shared_filenames": [ + "src/transformers/models/yolos/modeling_yolos.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/yolos/modeling_yolos.py", + "left_ranges": [ + [ + 446, + 452 + ] + ], + "right_ranges": [ + [ + 434, + 440 + ], + [ + 446, + 452 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45561-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45561.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45561, + "canonical_pr_number": 45645, + "issue_numbers": [ + 45561 + ], + "pr_numbers": [ + 45639, + 45645 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45639, + "right_pr_number": 45645, + "code_similarity": 0.721, + "size_similarity": 0.973, + "file_overlap": 0.75, + "area_overlap": 0.563, + "patch_similarity": 0.696, + "shared_filenames": [ + "src/transformers/testing_utils.py", + "tests/utils/test_testing_utils.py", + "utils/notification_service.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/testing_utils.py", + "left_ranges": [ + [ + 3611, + 3618 + ], + [ + 3835, + 3869 + ] + ], + "right_ranges": [ + [ + 3525, + 3558 + ], + [ + 3782, + 3788 + ] + ] + }, + { + "filename": "tests/utils/test_testing_utils.py", + "left_ranges": [ + [ + 1, + 144 + ] + ], + "right_ranges": [ + [ + 1, + 114 + ] + ] + }, + { + "filename": "utils/notification_service.py", + "left_ranges": [ + [ + 935, + 967 + ] + ], + "right_ranges": [ + [ + 935, + 967 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-47096-3", + "summary": "Cluster of 3 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47096, + "issue_numbers": [], + "pr_numbers": [ + 47096, + 47306, + 47367 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47096, + "right_pr_number": 47306, + "code_similarity": 0.169, + "size_similarity": 0.78, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.089, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47096, + "right_pr_number": 47367, + "code_similarity": 0.025, + "size_similarity": 0.097, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.04, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 47306, + "right_pr_number": 47367, + "code_similarity": 0.021, + "size_similarity": 0.076, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.041, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-46305-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46305.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 46305, + "canonical_pr_number": 47162, + "issue_numbers": [ + 46305 + ], + "pr_numbers": [ + 46310, + 47162 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46310, + "right_pr_number": 47162, + "code_similarity": 0.577, + "size_similarity": 0.338, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.062, + "shared_filenames": [ + "src/transformers/models/moonshine_streaming/modeling_moonshine_streaming.py", + "src/transformers/models/moonshine_streaming/modular_moonshine_streaming.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/moonshine_streaming/modeling_moonshine_streaming.py", + "left_ranges": [ + [ + 356, + 362 + ] + ], + "right_ranges": [ + [ + 356, + 362 + ] + ] + }, + { + "filename": "src/transformers/models/moonshine_streaming/modular_moonshine_streaming.py", + "left_ranges": [ + [ + 269, + 275 + ] + ], + "right_ranges": [ + [ + 269, + 275 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45081-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45081.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45081, + "canonical_pr_number": 45317, + "issue_numbers": [ + 45081 + ], + "pr_numbers": [ + 45086, + 45317 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45086, + "right_pr_number": 45317, + "code_similarity": 0.589, + "size_similarity": 0.34, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.136, + "shared_filenames": [ + "src/transformers/tokenization_utils_tokenizers.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/tokenization_utils_tokenizers.py", + "left_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ], + "right_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-47429-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #47429.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 47429, + "canonical_pr_number": 47434, + "issue_numbers": [ + 47429 + ], + "pr_numbers": [ + 47431, + 47434 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47431, + "right_pr_number": 47434, + "code_similarity": 1.0, + "size_similarity": 1.0, + "file_overlap": 1.0, + "area_overlap": 1.0, + "patch_similarity": 1.0, + "shared_filenames": [ + "setup.py", + "src/transformers/dependency_versions_table.py" + ], + "shared_file_areas": [ + { + "filename": "setup.py", + "left_ranges": [ + [ + 146, + 152 + ] + ], + "right_ranges": [ + [ + 146, + 152 + ] + ] + }, + { + "filename": "src/transformers/dependency_versions_table.py", + "left_ranges": [ + [ + 73, + 79 + ] + ], + "right_ranges": [ + [ + 73, + 79 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43698-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43698, + "canonical_pr_number": 43779, + "issue_numbers": [ + 43698 + ], + "pr_numbers": [ + 43779, + 43816 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43779, + "right_pr_number": 43816, + "code_similarity": 0.418, + "size_similarity": 0.538, + "file_overlap": 1.0, + "area_overlap": 0.02, + "patch_similarity": 0.02, + "shared_filenames": [ + "src/transformers/integrations/integration_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/integrations/integration_utils.py", + "left_ranges": [ + [ + 2227, + 2241 + ], + [ + 2303, + 2309 + ] + ], + "right_ranges": [ + [ + 2278, + 2291 + ], + [ + 2309, + 2322 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-47175-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47298, + "issue_numbers": [], + "pr_numbers": [ + 47175, + 47298 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47175, + "right_pr_number": 47298, + "code_similarity": 0.019, + "size_similarity": 0.043, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.067, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-47123-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47123, + "issue_numbers": [], + "pr_numbers": [ + 47123, + 47280 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47123, + "right_pr_number": 47280, + "code_similarity": 0.547, + "size_similarity": 0.522, + "file_overlap": 1.0, + "area_overlap": 0.368, + "patch_similarity": 0.093, + "shared_filenames": [ + "tests/models/minimax_m3_vl/test_modeling_minimax_m3_vl.py" + ], + "shared_file_areas": [ + { + "filename": "tests/models/minimax_m3_vl/test_modeling_minimax_m3_vl.py", + "left_ranges": [ + [ + 14, + 20 + ], + [ + 487, + 493 + ] + ], + "right_ranges": [ + [ + 14, + 21 + ], + [ + 481, + 499 + ], + [ + 520, + 530 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-47282-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47380, + "issue_numbers": [], + "pr_numbers": [ + 47282, + 47380 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47282, + "right_pr_number": 47380, + "code_similarity": 0.476, + "size_similarity": 0.5, + "file_overlap": 1.0, + "area_overlap": 0.194, + "patch_similarity": 0.056, + "shared_filenames": [ + "tests/models/mamba2/test_modeling_mamba2.py" + ], + "shared_file_areas": [ + { + "filename": "tests/models/mamba2/test_modeling_mamba2.py", + "left_ranges": [ + [ + 420, + 438 + ], + [ + 450, + 474 + ], + [ + 482, + 524 + ] + ], + "right_ranges": [ + [ + 18, + 24 + ], + [ + 406, + 419 + ], + [ + 426, + 432 + ], + [ + 457, + 463 + ], + [ + 487, + 493 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-47443-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 47444, + "issue_numbers": [], + "pr_numbers": [ + 47443, + 47444 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 47443, + "right_pr_number": 47444, + "code_similarity": 0.754, + "size_similarity": 0.632, + "file_overlap": 0.818, + "area_overlap": 1.0, + "patch_similarity": 0.217, + "shared_filenames": [ + "src/transformers/models/deepseek_v3/modeling_deepseek_v3.py", + "src/transformers/models/deepseek_v32/modeling_deepseek_v32.py", + "src/transformers/models/glm4_moe_lite/modeling_glm4_moe_lite.py", + "src/transformers/models/glm_moe_dsa/modeling_glm_moe_dsa.py", + "src/transformers/models/longcat_flash/modeling_longcat_flash.py", + "src/transformers/models/minicpm3/modeling_minicpm3.py", + "src/transformers/models/mistral4/modeling_mistral4.py", + "src/transformers/models/nemotron_h/modeling_nemotron_h.py", + "src/transformers/models/youtu/modeling_youtu.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/deepseek_v3/modeling_deepseek_v3.py", + "left_ranges": [ + [ + 4, + 9 + ], + [ + 310, + 315 + ] + ], + "right_ranges": [ + [ + 4, + 9 + ], + [ + 310, + 315 + ] + ] + }, + { + "filename": "src/transformers/models/deepseek_v32/modeling_deepseek_v32.py", + "left_ranges": [ + [ + 299, + 304 + ], + [ + 338, + 349 + ], + [ + 392, + 403 + ] + ], + "right_ranges": [ + [ + 299, + 304 + ], + [ + 338, + 349 + ], + [ + 392, + 403 + ] + ] + }, + { + "filename": "src/transformers/models/glm4_moe_lite/modeling_glm4_moe_lite.py", + "left_ranges": [ + [ + 182, + 187 + ], + [ + 221, + 232 + ], + [ + 272, + 283 + ] + ], + "right_ranges": [ + [ + 182, + 187 + ], + [ + 221, + 232 + ], + [ + 272, + 283 + ] + ] + }, + { + "filename": "src/transformers/models/glm_moe_dsa/modeling_glm_moe_dsa.py", + "left_ranges": [ + [ + 306, + 311 + ], + [ + 358, + 369 + ] + ], + "right_ranges": [ + [ + 306, + 311 + ], + [ + 358, + 369 + ] + ] + }, + { + "filename": "src/transformers/models/longcat_flash/modeling_longcat_flash.py", + "left_ranges": [ + [ + 282, + 287 + ], + [ + 321, + 332 + ], + [ + 372, + 383 + ] + ], + "right_ranges": [ + [ + 282, + 287 + ], + [ + 321, + 332 + ], + [ + 372, + 383 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modeling_minicpm3.py", + "left_ranges": [ + [ + 196, + 201 + ], + [ + 283, + 288 + ] + ], + "right_ranges": [ + [ + 196, + 201 + ], + [ + 283, + 288 + ] + ] + }, + { + "filename": "src/transformers/models/mistral4/modeling_mistral4.py", + "left_ranges": [ + [ + 17, + 22 + ], + [ + 319, + 324 + ] + ], + "right_ranges": [ + [ + 17, + 22 + ], + [ + 319, + 324 + ] + ] + }, + { + "filename": "src/transformers/models/nemotron_h/modeling_nemotron_h.py", + "left_ranges": [ + [ + 109, + 114 + ], + [ + 291, + 300 + ], + [ + 361, + 370 + ], + [ + 396, + 409 + ], + [ + 435, + 444 + ] + ], + "right_ranges": [ + [ + 109, + 114 + ], + [ + 291, + 300 + ], + [ + 361, + 370 + ], + [ + 396, + 409 + ], + [ + 435, + 444 + ] + ] + }, + { + "filename": "src/transformers/models/youtu/modeling_youtu.py", + "left_ranges": [ + [ + 222, + 227 + ], + [ + 261, + 272 + ], + [ + 312, + 323 + ] + ], + "right_ranges": [ + [ + 222, + 227 + ], + [ + 261, + 272 + ], + [ + 312, + 323 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-44018-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 44068, + "issue_numbers": [], + "pr_numbers": [ + 44018, + 44068 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 44018, + "right_pr_number": 44068, + "code_similarity": 0.766, + "size_similarity": 0.939, + "file_overlap": 1.0, + "area_overlap": 0.425, + "patch_similarity": 0.866, + "shared_filenames": [ + "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "tests/models/gpt_neo/test_modeling_gpt_neo.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "left_ranges": [ + [ + 26, + 46 + ], + [ + 138, + 143 + ], + [ + 182, + 187 + ], + [ + 283, + 295 + ], + [ + 327, + 341 + ], + [ + 348, + 354 + ], + [ + 360, + 369 + ], + [ + 399, + 406 + ], + [ + 411, + 419 + ], + [ + 428, + 433 + ], + [ + 472, + 492 + ], + [ + 507, + 513 + ], + [ + 519, + 528 + ], + [ + 541, + 559 + ], + [ + 562, + 567 + ], + [ + 595, + 601 + ], + [ + 607, + 614 + ], + [ + 627, + 643 + ], + [ + 685, + 690 + ], + [ + 708, + 714 + ], + [ + 720, + 727 + ], + [ + 740, + 757 + ], + [ + 761, + 766 + ], + [ + 780, + 786 + ], + [ + 791, + 798 + ], + [ + 807, + 822 + ], + [ + 840, + 845 + ] + ], + "right_ranges": [ + [ + 26, + 31 + ], + [ + 34, + 44 + ], + [ + 136, + 141 + ], + [ + 180, + 185 + ], + [ + 281, + 293 + ], + [ + 325, + 339 + ], + [ + 346, + 352 + ], + [ + 358, + 367 + ], + [ + 397, + 404 + ], + [ + 409, + 417 + ], + [ + 426, + 434 + ], + [ + 463, + 483 + ], + [ + 498, + 504 + ], + [ + 510, + 519 + ], + [ + 532, + 550 + ], + [ + 553, + 564 + ], + [ + 586, + 592 + ], + [ + 598, + 605 + ], + [ + 618, + 634 + ], + [ + 676, + 688 + ], + [ + 699, + 705 + ], + [ + 711, + 718 + ], + [ + 731, + 748 + ], + [ + 752, + 762 + ], + [ + 771, + 777 + ], + [ + 782, + 789 + ], + [ + 798, + 813 + ], + [ + 831, + 836 + ] + ] + }, + { + "filename": "tests/models/gpt_neo/test_modeling_gpt_neo.py", + "left_ranges": [ + [ + 458, + 464 + ] + ], + "right_ranges": [ + [ + 458, + 464 + ] + ] + } + ] + } + ] + } + ], + "duplicate_issues": [ + { + "cluster_id": "cluster-46894-7", + "canonical_issue_number": 46894, + "duplicate_issue_numbers": [ + 46897, + 46901 + ], + "reason": "Issues in cluster-46894-7 are treated as duplicates because they share closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47138-4", + "canonical_issue_number": 47138, + "duplicate_issue_numbers": [ + 47139 + ], + "reason": "Issues in cluster-47138-4 are treated as duplicates because they share closing_reference, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47332-4", + "canonical_issue_number": 47333, + "duplicate_issue_numbers": [ + 47332 + ], + "reason": "Issues in cluster-47332-4 are treated as duplicates because they share closing_reference, soft_similarity evidence." + } + ], + "duplicate_prs": [ + { + "cluster_id": "cluster-41211-3", + "canonical_pr_number": 44339, + "duplicate_pr_numbers": [ + 41356 + ], + "target_issue_number": 41211, + "reason": "PRs in cluster-41211-3 are treated as duplicates because they converge on issue #41211 with closing_reference evidence." + }, + { + "cluster_id": "cluster-43240-3", + "canonical_pr_number": 43251, + "duplicate_pr_numbers": [ + 43254 + ], + "target_issue_number": 43240, + "reason": "PRs in cluster-43240-3 are treated as duplicates because they converge on issue #43240 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43451-3", + "canonical_pr_number": 47322, + "duplicate_pr_numbers": [ + 43451 + ], + "target_issue_number": 47321, + "reason": "PRs in cluster-43451-3 are treated as duplicates because they converge on issue #47321 with closing_reference, soft_similarity evidence." + }, + { + "cluster_id": "cluster-43698-3", + "canonical_pr_number": 43779, + "duplicate_pr_numbers": [ + 43816 + ], + "target_issue_number": 43698, + "reason": "PRs in cluster-43698-3 are treated as duplicates because they converge on issue #43698 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43979-11", + "canonical_pr_number": 44007, + "duplicate_pr_numbers": [ + 43996, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "target_issue_number": 43979, + "reason": "PRs in cluster-43979-11 are treated as duplicates because they converge on issue #43979 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-44018-2", + "canonical_pr_number": 44068, + "duplicate_pr_numbers": [ + 44018 + ], + "target_issue_number": null, + "reason": "PRs in cluster-44018-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-45081-3", + "canonical_pr_number": 45317, + "duplicate_pr_numbers": [ + 45086 + ], + "target_issue_number": 45081, + "reason": "PRs in cluster-45081-3 are treated as duplicates because they converge on issue #45081 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45561-3", + "canonical_pr_number": 45645, + "duplicate_pr_numbers": [ + 45639 + ], + "target_issue_number": 45561, + "reason": "PRs in cluster-45561-3 are treated as duplicates because they converge on issue #45561 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45647-3", + "canonical_pr_number": 45738, + "duplicate_pr_numbers": [ + 45737 + ], + "target_issue_number": 45647, + "reason": "PRs in cluster-45647-3 are treated as duplicates because they converge on issue #45647 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-46159-3", + "canonical_pr_number": 46174, + "duplicate_pr_numbers": [ + 46780 + ], + "target_issue_number": 46159, + "reason": "PRs in cluster-46159-3 are treated as duplicates because they converge on issue #46159 with closing_reference evidence." + }, + { + "cluster_id": "cluster-46305-3", + "canonical_pr_number": 47162, + "duplicate_pr_numbers": [ + 46310 + ], + "target_issue_number": 46305, + "reason": "PRs in cluster-46305-3 are treated as duplicates because they converge on issue #46305 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-46509-5", + "canonical_pr_number": 47143, + "duplicate_pr_numbers": [ + 46509, + 47163, + 47301, + 47387 + ], + "target_issue_number": null, + "reason": "PRs in cluster-46509-5 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-46869-3", + "canonical_pr_number": 46886, + "duplicate_pr_numbers": [ + 46922 + ], + "target_issue_number": 46869, + "reason": "PRs in cluster-46869-3 are treated as duplicates because they converge on issue #46869 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-46894-7", + "canonical_pr_number": 46903, + "duplicate_pr_numbers": [ + 46895, + 46958, + 47090 + ], + "target_issue_number": 46894, + "reason": "PRs in cluster-46894-7 are treated as duplicates because they converge on issue #46894 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47096-3", + "canonical_pr_number": 47096, + "duplicate_pr_numbers": [ + 47306, + 47367 + ], + "target_issue_number": null, + "reason": "PRs in cluster-47096-3 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-47123-2", + "canonical_pr_number": 47123, + "duplicate_pr_numbers": [ + 47280 + ], + "target_issue_number": null, + "reason": "PRs in cluster-47123-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-47138-4", + "canonical_pr_number": 47303, + "duplicate_pr_numbers": [ + 47140 + ], + "target_issue_number": 47138, + "reason": "PRs in cluster-47138-4 are treated as duplicates because they converge on issue #47138 with closing_reference, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47175-2", + "canonical_pr_number": 47298, + "duplicate_pr_numbers": [ + 47175 + ], + "target_issue_number": null, + "reason": "PRs in cluster-47175-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-47282-2", + "canonical_pr_number": 47380, + "duplicate_pr_numbers": [ + 47282 + ], + "target_issue_number": null, + "reason": "PRs in cluster-47282-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-47332-4", + "canonical_pr_number": 47336, + "duplicate_pr_numbers": [ + 47383 + ], + "target_issue_number": 47333, + "reason": "PRs in cluster-47332-4 are treated as duplicates because they converge on issue #47333 with closing_reference, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47429-3", + "canonical_pr_number": 47434, + "duplicate_pr_numbers": [ + 47431 + ], + "target_issue_number": 47429, + "reason": "PRs in cluster-47429-3 are treated as duplicates because they converge on issue #47429 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-47443-2", + "canonical_pr_number": 47444, + "duplicate_pr_numbers": [ + 47443 + ], + "target_issue_number": null, + "reason": "PRs in cluster-47443-2 are treated as duplicates because they share soft_similarity evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-46894-7", + "issue_number": 46894, + "reason": "Issue #46894 is the strongest global issue candidate because it is open, belongs to a cluster with 7 artifacts, and carries 1 discussion comments plus 10 inbound references.", + "score": 182.0 + }, + "best_pr": { + "cluster_id": "cluster-43979-11", + "pr_number": 44007, + "reason": "PR #44007 is the strongest global PR candidate because it is open, belongs to a cluster with 11 artifacts, links to 1 issues, and carries 2 review/discussion events.", + "score": 238.886 + } +} diff --git a/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..fdabdef7912f35457f37d6186120b6c4c5941688 --- /dev/null +++ b/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,29368 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260721T060018Z", + "generated_at": "2026-07-21T06:15:09Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:24643", + "issue:30277", + "issue:34634", + "issue:34928", + "issue:35308", + "issue:35545", + "issue:36010", + "issue:36331", + "issue:39290", + "issue:41720", + "issue:41762", + "issue:42915", + "issue:43039", + "issue:43425", + "issue:43541", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43828", + "issue:43975", + "issue:44292", + "issue:44322", + "issue:44457", + "issue:44509", + "issue:44512", + "issue:44560", + "issue:44805", + "issue:44841", + "issue:44918", + "issue:45081", + "issue:45161", + "issue:45237", + "issue:45464", + "issue:45507", + "issue:45685", + "issue:45800", + "issue:45823", + "issue:45859", + "issue:45874", + "issue:45901", + "issue:45902", + "issue:46015", + "issue:46165", + "issue:46294", + "issue:46468", + "issue:46515", + "issue:46661", + "issue:46869" + ], + "soft_pairs": [ + "issue:44322|issue:45464", + "issue:45237|issue:45685", + "issue:43716|issue:45237", + "issue:43828|issue:45237", + "issue:45237|issue:46468", + "issue:45237|issue:45507", + "issue:30277|issue:35545", + "issue:35545|issue:42915", + "issue:35545|issue:41720", + "issue:45901|issue:46015", + "issue:35545|issue:36010", + "issue:44918|issue:46165", + "issue:46165|issue:46869", + "issue:44509|issue:44512", + "issue:44805|issue:45823", + "issue:43638|issue:44805", + "issue:43638|issue:46165", + "issue:43596|issue:44805", + "issue:44805|issue:46294", + "issue:46165|issue:46468", + "issue:44805|issue:45161", + "issue:44560|issue:46165", + "issue:34634|issue:35545", + "issue:30277|issue:43638", + "issue:43596|issue:43638", + "issue:43039|issue:45161", + "issue:43975|issue:44457", + "issue:34928|issue:35308", + "issue:44292|issue:44457", + "issue:44841|issue:46661", + "issue:24643|issue:30277", + "issue:43716|issue:46869", + "issue:34928|issue:44805", + "issue:45685|issue:46869", + "issue:34928|issue:36331", + "issue:45859|issue:45902", + "issue:43638|issue:45161", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:44292|issue:45237", + "issue:45874|issue:46294", + "issue:39290|issue:45081", + "issue:45800|issue:46515", + "issue:34928|issue:41720", + "issue:45823|issue:45874", + "issue:43425|issue:44292", + "issue:43828|issue:45161", + "issue:43541|issue:43716", + "issue:43828|issue:44292", + "issue:43716|issue:45685", + "issue:45507|issue:45685", + "issue:44292|issue:44918", + "issue:43716|issue:44292", + "issue:44918|issue:45161", + "issue:43039|issue:44918", + "issue:41762|issue:43638" + ], + "prepared_review_unit_hash": "1b6f1566d5de3821abc767cf96c50cf01873b44348a87cfc903e0678e17740c8", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 56, + "serialized_chars": 44520, + "estimated_input_tokens": 11130, + "estimated_eval_tokens": 22516 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous training/model bugs, with two credible duplicate pockets: ZeRO-3 BERT init (43596/43638) and Qwen3_5Moe rotary embedding materialization (45859/45902). The torch 2.4.1 / transformers pair (45800/46515) is too weakly evidenced to call an exact duplicate.", + "confidence": 0.71, + "canonical_issue_reason": "Issue 43638 is the strongest exact-duplicate anchor: 43596 describes the same ZeRO-3 BERT empty-tensor / index-out-of-bounds initialization failure.", + "canonical_pr_reason": null, + "best_issue_reason": "43638 is the best global issue representative because it anchors the clearest confirmed duplicate pocket, while most of the surrounding cluster is unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Both involve Qwen3.5, but one is a missing config attribute and the other is a streaming chat/completions failure; different bugs." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Different models and failure modes: GPT-OSS AMD GPU support vs an MPS histogram kernel error." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Different model stacks and symptoms; a dtype mismatch in Mistral-3 is not the same bug as GPT-OSS failing on AMD GPUs." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Phi-tiny-MoE autocast dtype mismatch and GPT-OSS AMD GPU failure are unrelated." + }, + { + "left": "issue:45237", + "right": "issue:46468", + "accept": false, + "reason": "Both are runtime failures, but they concern different models and code paths, not the same underlying bug." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "GPT-OSS AMD GPU failure and GraniteMoEHybrid invalid method call are unrelated." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed ZeRO3 collective mismatch is unrelated to a ModernBERT ONNX export error." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ONNX export failure and FineGrainedFP8Config failure are different model/runtime problems." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "ModernBERT ONNX export and Qwen3 auto device map cuda assert are unrelated failures." + }, + { + "left": "issue:45901", + "right": "issue:46015", + "accept": false, + "reason": "Table QA crash and outdated RAG docs are different issue types and different fixes." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "ONNX export error is unrelated to a missing GenerationMixin import." + }, + { + "left": "issue:44918", + "right": "issue:46165", + "accept": false, + "reason": "Qwen3.5 embedding unpacking under TRL SFT trainer is not the same as resume-training with trust_remote_code." + }, + { + "left": "issue:46165", + "right": "issue:46869", + "accept": false, + "reason": "Resume-training trust_remote_code and YOLO multi-GPU device_map issues are unrelated." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are v5 docs regressions, but they reference different removed docs entries and different cleanup work." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Mask shape mismatch is unrelated to Gemma4 PLE device mismatch." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "ZeRO-3 BERT init failure and a mask/indexing shape mismatch are different bugs." + }, + { + "left": "issue:43638", + "right": "issue:46165", + "accept": false, + "reason": "ZeRO-3 BERT init failure is unrelated to the resume-training trust_remote_code issue." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Both are tensor/indexing errors, but the affected models and code paths differ." + }, + { + "left": "issue:44805", + "right": "issue:46294", + "accept": false, + "reason": "Mask/index shape mismatch is not the same as Qwen3.5 video token / feature mismatch." + }, + { + "left": "issue:46165", + "right": "issue:46468", + "accept": false, + "reason": "Resume-training with remote code and a broken Nemotron example are unrelated." + }, + { + "left": "issue:44560", + "right": "issue:46165", + "accept": false, + "reason": "Qwen3-vl video StopIteration and resume-training trust_remote_code are different bugs." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "BarkProcessor voice_preset and ModernBERT ONNX export are unrelated." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Both involve DeepSpeed, but the concrete failures are different and not mergeable into one fix." + }, + { + "left": "issue:43716", + "right": "issue:46869", + "accept": false, + "reason": "Mistral-3 dtype mismatch and YOLO multi-GPU device map failure are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "FSDP activation-checkpointing mismatch and mask indexing shape mismatch are different distributed-training bugs." + }, + { + "left": "issue:45685", + "right": "issue:46869", + "accept": false, + "reason": "MPS histogram kernel error and YOLO device_map auto failure are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "FSDP activation-checkpointing mismatch is unrelated to a Trainer API signature change." + }, + { + "left": "issue:45859", + "right": "issue:45902", + "accept": true, + "reason": "Same Qwen3_5MoeTextRotaryEmbedding path; both indicate broken device/materialization handling of rotary embedding state under offload/meta-to-empty flows." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "ZeRO-3 BERT init failure and GPT-OSS tensor-parallel failure are different bugs." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "Same as above: different model and failure mode." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Mistral-3 dtype mismatch and GraniteMoEHybrid invalid method call are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Qwen NVFP4 runtime error and GPT-OSS AMD GPU failure are unrelated." + }, + { + "left": "issue:45874", + "right": "issue:46294", + "accept": false, + "reason": "Gemma4 reverse embedding lookup and Qwen3.5 video token mismatch are different bugs." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Gemma3TextConfig missing attribute and Mistral tokenizer regex patch crash are unrelated." + }, + { + "left": "issue:45800", + "right": "issue:46515", + "accept": false, + "reason": "The titles are similar, but 46515 only says torch 2.4.1 incompatibility and does not establish the same transformers 5.8.0-specific bug or fix path." + }, + { + "left": "issue:34928", + "right": "issue:41720", + "accept": false, + "reason": "FSDP activation checkpointing mismatch and Qwen3 cuda assert are different failures." + }, + { + "left": "issue:45823", + "right": "issue:45874", + "accept": false, + "reason": "Gemma4 PLE device mismatch and Gemma4 inputs_embeds reverse lookup are separate issues." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch 2.10 incompatibility and Qwen-3-8B-NVFP4 error are different version/model issues." + }, + { + "left": "issue:43828", + "right": "issue:45161", + "accept": false, + "reason": "Phi-tiny-MoE autocast dtype mismatch is unrelated to GPT-OSS tensor-parallel failure." + }, + { + "left": "issue:43541", + "right": "issue:43716", + "accept": false, + "reason": "Mixtral grouped_mm torch dynamo tracing failure and Mistral-3 dtype mismatch are different bugs." + }, + { + "left": "issue:43828", + "right": "issue:44292", + "accept": false, + "reason": "Different models and failures: autocast dtype mismatch vs NVFP4 runtime error." + }, + { + "left": "issue:43716", + "right": "issue:45685", + "accept": false, + "reason": "Mistral-3 image preprocessor dtype mismatch and MPS histogram kernel error are unrelated." + }, + { + "left": "issue:45507", + "right": "issue:45685", + "accept": false, + "reason": "GraniteMoEHybrid invalid method and MPS histogram kernel error are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:44918", + "accept": false, + "reason": "Qwen NVFP4 runtime error and TRL SFT embedding unpacking are different issues." + }, + { + "left": "issue:43039", + "right": "issue:44918", + "accept": false, + "reason": "Liger Kernel cross_entropy usage and Qwen3.5 embedding unpacking are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": false, + "reason": "Gemma3 ZeRO-3 load failure and BERT ZeRO-3 init failure are only broadly similar; the concrete bugs differ." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The overall clustering read is conservative and mostly grounded, but one soft-edge accept is too strong for the packet. In particular, issue:45800 vs issue:46515 is described as an exact duplicate even though the packet only shows similar incompatibility titles; 46515 does not explicitly mention transformers 5.8.0, so 'exact same' overstates the evidence. The Qwen3.5 rotary-embedding pair is more defensible as a likely duplicate, and most other rejections look appropriately cautious, but the overconfident 45800/46515 verdict means I would not accept this review as-is." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 2, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:30277", + "issue:34928", + "issue:35308", + "issue:36010", + "issue:36331", + "issue:39290", + "issue:41093", + "issue:41720", + "issue:41762", + "issue:42617", + "issue:43039", + "issue:43296", + "issue:43425", + "issue:43531", + "issue:43596", + "issue:43638", + "issue:43792", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43872", + "issue:43901", + "issue:43975", + "issue:44292", + "issue:44512", + "issue:44560", + "issue:44589", + "issue:44805", + "issue:44841", + "issue:44863", + "issue:45020", + "issue:45070", + "issue:45237", + "issue:45685", + "issue:45800", + "issue:45812", + "issue:45823", + "issue:45901", + "issue:45920", + "issue:46015", + "issue:46294", + "issue:46304", + "issue:46402", + "issue:46772", + "issue:46901", + "issue:47394" + ], + "soft_pairs": [ + "issue:43827|issue:44512", + "issue:43828|issue:45685", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:34928|issue:36010", + "issue:45237|issue:46015", + "issue:43039|issue:44805", + "issue:45800|issue:46304", + "issue:43296|issue:46304", + "issue:39290|issue:41720", + "issue:43866|issue:46402", + "issue:44589|issue:46304", + "issue:45237|issue:45901", + "issue:41762|issue:43596", + "issue:43039|issue:44560", + "issue:43901|issue:44512", + "issue:44841|issue:45823", + "issue:45823|issue:46294", + "issue:43975|issue:46015", + "issue:45020|issue:46304", + "issue:43866|issue:46772", + "issue:44863|issue:46015", + "issue:35308|issue:42617", + "issue:45812|issue:45920", + "issue:43872|issue:46304", + "issue:39290|issue:43531", + "issue:44292|issue:46015", + "issue:46901|issue:47394", + "issue:43866|issue:44863", + "issue:41093|issue:43638", + "issue:43792|issue:45901", + "issue:43425|issue:45070", + "issue:43854|issue:43866" + ], + "prepared_review_unit_hash": "acdfc73bf66d72ebcab8bdf0dfae14b3bb4f794f2178ea915420c5f575e91b64", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 36962, + "estimated_input_tokens": 9241, + "estimated_eval_tokens": 18738 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The items split into mostly unrelated issues, with two clear duplicate subclusters: a v5 AutoTokenizer wrong-token-ID regression and a DeepSpeed ZeRO-3 zero-size loading/index error. I only accepted soft edges where the titles describe the same concrete bug; most others are broad subsystem similarities, not true duplicates.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45920 is the broadest and most representative report of the tokenizer regression, covering multiple models rather than a single Granite-only case.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45920 best captures the shared underlying v5 AutoTokenizer bug and is the strongest canonical issue for the accepted duplicate family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs bugs, but they concern different removed commands and different pages." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Unrelated runtime errors: autocast dtype mismatch vs MPS histogram_int limitation." + }, + { + "left": "issue:43039", + "right": "issue:43828", + "accept": false, + "reason": "Different bugs in different code paths: Liger cross-entropy vs autocast dtype mismatch." + }, + { + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "Different failures: collective mismatch under ZeRO-3 vs trainer API kwarg mismatch." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "Activation-checkpoint/FSDP size mismatch and GenerationMixin import error are unrelated." + }, + { + "left": "issue:45237", + "right": "issue:46015", + "accept": false, + "reason": "GPU backend support issue vs documentation maintenance; not the same bug." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Different concrete failures: kernel call behavior vs mask/tensor shape mismatch." + }, + { + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Both mention version compatibility, but the affected environment and failure mode differ." + }, + { + "left": "issue:43296", + "right": "issue:46304", + "accept": false, + "reason": "vLLM model loading failure is unrelated to the distributed-support regression." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Different models and different symptoms: missing config attr vs CUDA assert." + }, + { + "left": "issue:43866", + "right": "issue:46402", + "accept": false, + "reason": "Checkpoint corruption and weights-not-loaded are distinct model-loading problems." + }, + { + "left": "issue:44589", + "right": "issue:46304", + "accept": false, + "reason": "Storage-class lookup failure is unrelated to USE_DISTRIBUTED=0 support." + }, + { + "left": "issue:45237", + "right": "issue:45901", + "accept": false, + "reason": "AMD GPU incompatibility and table QA crash are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": true, + "reason": "Both describe the same ZeRO-3 loading bug leading to zero-sized tensors and index-out-of-bounds failures." + }, + { + "left": "issue:43039", + "right": "issue:44560", + "accept": false, + "reason": "Cross-entropy kernel behavior and video StopIteration are unrelated." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Different documentation mismatches about different APIs/commands." + }, + { + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Processor loading failure and device_map forward mismatch are different bugs." + }, + { + "left": "issue:45823", + "right": "issue:46294", + "accept": false, + "reason": "Device placement error and chat-template video-token mismatch are different concrete issues." + }, + { + "left": "issue:43975", + "right": "issue:46015", + "accept": false, + "reason": "Detokenization bug and docs cleanup are unrelated." + }, + { + "left": "issue:45020", + "right": "issue:46304", + "accept": false, + "reason": "Remote-code breakage and distributed-build support regression are separate compatibility issues." + }, + { + "left": "issue:43866", + "right": "issue:46772", + "accept": false, + "reason": "Checkpoint corruption is unrelated to NVFP4 compatibility." + }, + { + "left": "issue:44863", + "right": "issue:46015", + "accept": false, + "reason": "Model checkpoint loading failure vs outdated examples are unrelated." + }, + { + "left": "issue:35308", + "right": "issue:42617", + "accept": false, + "reason": "Both are distributed-training related, but the failure modes and root causes differ." + }, + { + "left": "issue:45812", + "right": "issue:45920", + "accept": true, + "reason": "Same AutoTokenizer wrong-token-ID regression; 45920 is just the broader multi-model report." + }, + { + "left": "issue:43872", + "right": "issue:46304", + "accept": false, + "reason": "bitsandbytes kwarg incompatibility is unrelated to the distributed-support issue." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Both mention sliding_window, but one is a missing config field and the other is a different model behavior issue." + }, + { + "left": "issue:44292", + "right": "issue:46015", + "accept": false, + "reason": "NVFP4 runtime failure and docs issues are unrelated." + }, + { + "left": "issue:46901", + "right": "issue:47394", + "accept": false, + "reason": "Similar loss-shift theme, but different models and the second adds cross-row leakage; not clearly the same code path." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Different model-loading failures with different symptoms and likely different causes." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "The error strings differ and the reports point to different failure modes, despite both involving ZeRO-3." + }, + { + "left": "issue:43792", + "right": "issue:45901", + "accept": false, + "reason": "Whisper load failure and table-question-answering crash are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch/transformers version incompatibility and pydantic config-field breakage are different regressions." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Unit-test load failure and checkpoint corruption are distinct problems." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays grounded in the packet and the soft-edge judgments are conservative. The two accepted duplicate pairs are well supported by the issue titles: 45812/45920 are both AutoTokenizer wrong-token-ID regressions, and 41762/43596 are both DeepSpeed ZeRO-3 zero-size/index errors. The remaining rejections avoid overstating similarity and mostly distinguish different concrete failure modes or different docs pages." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 3, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:36010", + "issue:39186", + "issue:39290", + "issue:41093", + "issue:41762", + "issue:43054", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43716", + "issue:43792", + "issue:43828", + "issue:43866", + "issue:43872", + "issue:43975", + "issue:43994", + "issue:44291", + "issue:44292", + "issue:44387", + "issue:44479", + "issue:44805", + "issue:44841", + "issue:44918", + "issue:45005", + "issue:45020", + "issue:45070", + "issue:45084", + "issue:45362", + "issue:45399", + "issue:45464", + "issue:45507", + "issue:45685", + "issue:45759", + "issue:45800", + "issue:45823", + "issue:45901", + "issue:45902", + "issue:46165", + "issue:46294", + "issue:46468", + "issue:46640", + "issue:46661", + "issue:46869", + "issue:47228" + ], + "soft_pairs": [ + "issue:43541|issue:47228", + "issue:43792|issue:45070", + "issue:43596|issue:45800", + "issue:43975|issue:45901", + "issue:43541|issue:46869", + "issue:45362|issue:45464", + "issue:44292|issue:45901", + "issue:39290|issue:45823", + "issue:36010|issue:39290", + "issue:45464|issue:45685", + "issue:43541|issue:45685", + "issue:43541|issue:43828", + "issue:44841|issue:45084", + "issue:45507|issue:46869", + "issue:43404|issue:43866", + "issue:45399|issue:46661", + "issue:43572|issue:45070", + "issue:41093|issue:41762", + "issue:45800|issue:45823", + "issue:43872|issue:47228", + "issue:44805|issue:44918", + "issue:43716|issue:47228", + "issue:43828|issue:45464", + "issue:43296|issue:43366", + "issue:43366|issue:45464", + "issue:44291|issue:45070", + "issue:43531|issue:45070", + "issue:46869|issue:47228", + "issue:44387|issue:45005", + "issue:45020|issue:45800", + "issue:45507|issue:47228", + "issue:43716|issue:46294", + "issue:44479|issue:46294", + "issue:41093|issue:43596", + "issue:43054|issue:43994", + "issue:46468|issue:46869", + "issue:45759|issue:45902", + "issue:46640|issue:46661", + "issue:43596|issue:46165", + "issue:39186|issue:44918", + "issue:43716|issue:44918" + ], + "prepared_review_unit_hash": "01f16fb93b739e733167088262eb9746d9f1c19c011e73d14a69dc137c12e8f3", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 41, + "serialized_chars": 39844, + "estimated_input_tokens": 9961, + "estimated_eval_tokens": 20178 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly noisy and cross-cutting: most pairs share only broad keywords like dtype, index, or Qwen/Gemma without the same code path. Only one soft edge looks like a true duplicate-level match: the Qwen3.5 video-input regression pair (44479/46294).", + "confidence": 0.74, + "canonical_issue_reason": "Issue 44479 is the best anchor here because it describes a concrete, reproducible Qwen multimodal regression and has the strongest duplicate-like follow-up in the supplied pairs.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44479 is the most suitable global issue anchor: it is specific, actionable, and closest to a genuine same-bug cluster among the candidates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43541", + "right": "issue:47228", + "accept": false, + "reason": "Different model and failure mode: Mixtral torch-dynamo grouped_mm tracing vs Sam3 bf16 compilation regression." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper load failure vs a PretrainedConfig/pydantic compatibility break; unrelated paths." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "BertModel ZeRO-3 index error vs a broad torch/transformers version incompatibility report." + }, + { + "left": "issue:43975", + "right": "issue:45901", + "accept": false, + "reason": "Detokenization bug in v5 vs table-question-answering crash; different subsystems." + }, + { + "left": "issue:43541", + "right": "issue:46869", + "accept": false, + "reason": "Mixtral compile/tracing issue vs YOLO multi-GPU device-map failure." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Same general Qwen3.5 family, but different stacks and paths: Transformers chat vs vLLM streaming API." + }, + { + "left": "issue:44292", + "right": "issue:45901", + "accept": false, + "reason": "Qwen NVFP4 load failure vs table QA crash; no shared bug." + }, + { + "left": "issue:39290", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3 sliding-window config error vs Gemma4 device placement mismatch." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "Missing GenerationMixin import and Gemma3 sliding_window_pattern are unrelated issues." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Qwen3.5 streaming inference failure vs MPS histogram op limitation; different code paths." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Mixtral compile bug vs MPS int histogram limitation." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Torch-dynamo grouped_mm failure vs autocast dtype mismatch; not the same bug." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Voxtral processor failure vs template compilation error; unrelated." + }, + { + "left": "issue:45507", + "right": "issue:46869", + "accept": false, + "reason": "GraniteMoE invalid method vs YOLO multi-GPU device-map issue." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Tied lm_head bug in Mistral3 vs corrupted Ovis2 checkpoint." + }, + { + "left": "issue:45399", + "right": "issue:46661", + "accept": false, + "reason": "flash-attn2 fallback gating vs remote-code model config_class error." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "StableLm pad_token_idx regression vs generic PretrainedConfig field break; too broad to merge." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Both are index/shape errors, but one is mask-length mismatch and the other is DeepSpeed ZeRO-3 empty-load failure." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Torch/transformers incompatibility vs Gemma4 device-map mismatch; not the same underlying bug." + }, + { + "left": "issue:43872", + "right": "issue:47228", + "accept": false, + "reason": "bitsandbytes constructor incompatibility vs Sam3 bf16 compilation regression." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Similar shape-error wording, but one is a mask/index mismatch and the other is TRL embedding unpacking." + }, + { + "left": "issue:43716", + "right": "issue:47228", + "accept": false, + "reason": "Mistral image-preprocessor dtype mismatch vs Sam3 compilation regression." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Autocast dtype mismatch in Phi-tiny-MoE vs Qwen3.5 streaming API failure." + }, + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "PaddleOCR-VL load problem in vLLM vs GGUF gpt-oss architecture support." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "GGUF architecture support vs Qwen chat/streaming failure; unrelated." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized constructor issue vs PretrainedConfig pydantic regression." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Qwen3-MoE sliding_window bug vs generic config-field breakage." + }, + { + "left": "issue:46869", + "right": "issue:47228", + "accept": false, + "reason": "YOLO device-map problem vs Sam3 bf16 compilation regression." + }, + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "int4 memory/OOM regression vs tied-weights translation-model bug." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "Remote-code breakage and torch/transformers incompatibility are both broad version issues, but not the same concrete bug." + }, + { + "left": "issue:45507", + "right": "issue:47228", + "accept": false, + "reason": "GraniteMoE invalid method vs Sam3 compilation regression." + }, + { + "left": "issue:43716", + "right": "issue:46294", + "accept": false, + "reason": "Both are multimodal input mismatches, but image-preprocessor dtype failure and video token/feature mismatch are different bugs." + }, + { + "left": "issue:44479", + "right": "issue:46294", + "accept": true, + "reason": "Same Qwen video-input regression family with the same apply_chat_template/video-token mismatch theme; plausible same bug." + }, + { + "left": "issue:46468", + "right": "issue:46869", + "accept": false, + "reason": "Nemotron example failure vs YOLO device-map issue; unrelated." + }, + { + "left": "issue:45759", + "right": "issue:45902", + "accept": false, + "reason": "Composite-config unwrapping bug vs uninitialized-memory rotary embedding bug in Qwen3.5 MoE." + }, + { + "left": "issue:46640", + "right": "issue:46661", + "accept": false, + "reason": "Flash Attention 2 issue vs remote-code model loading error." + }, + { + "left": "issue:43596", + "right": "issue:46165", + "accept": false, + "reason": "ZeRO-3 BertModel load failure vs resume-with-remote-code training issue." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP 2-D weight error vs TRL embedding unpacking error." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Mistral multimodal dtype issue vs Qwen3.5 SFT unpacking issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The summary stays within the packet, and the only accepted soft edge (44479/46294) is plausibly the same Qwen video-input regression family based on the titles. The rejections also avoid overclaiming shared bugs from broad keyword overlap." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 4, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:33666", + "issue:35308", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:40990", + "issue:41762", + "issue:42915", + "issue:43054", + "issue:43404", + "issue:43644", + "issue:43645", + "issue:43828", + "issue:43872", + "issue:43873", + "issue:43950", + "issue:43975", + "issue:44292", + "issue:44360", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44534", + "issue:44589", + "issue:44863", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:45005", + "issue:45183", + "issue:45399", + "issue:45507", + "issue:45632", + "issue:45698", + "issue:45701", + "issue:45902", + "issue:45920", + "issue:45987", + "issue:46032", + "issue:46463", + "issue:46489", + "issue:46491", + "issue:46858", + "issue:46899", + "issue:46932", + "issue:47246", + "issue:47332", + "issue:47333" + ], + "soft_pairs": [ + "issue:43645|issue:45399", + "issue:40990|issue:44960", + "issue:42915|issue:45399", + "issue:43404|issue:45005", + "issue:45632|issue:45698", + "issue:43873|issue:45005", + "issue:35308|issue:36331", + "issue:44360|issue:44512", + "issue:44492|issue:44512", + "issue:44918|issue:45507", + "issue:44530|issue:45005", + "issue:43975|issue:44292", + "issue:45183|issue:46463", + "issue:35308|issue:42915", + "issue:44863|issue:45005", + "issue:39692|issue:43054", + "issue:43950|issue:44534", + "issue:46489|issue:46491", + "issue:36331|issue:41762", + "issue:43644|issue:43950", + "issue:33666|issue:35308", + "issue:46032|issue:47246", + "issue:43872|issue:44589", + "issue:46899|issue:46932", + "issue:43054|issue:45701", + "issue:45920|issue:46489", + "issue:47332|issue:47333", + "issue:38175|issue:43054", + "issue:43828|issue:44928", + "issue:45987|issue:46858", + "issue:43950|issue:45902" + ], + "prepared_review_unit_hash": "6cf6cbe577800ab6e6cdf579e4bff21e4fd6900503b160778ed04b4058616a6f", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37135, + "estimated_input_tokens": 9284, + "estimated_eval_tokens": 18824 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a mix of unrelated issues, with a few clear duplicate families around non-persistent buffer initialization, cache sequence-length return types, offloading save failures, and a couple of model-specific regressions.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43644 is the clearest root-cause report in the non-persistent-buffer regression family: it states the broad bug plainly and matches nearby duplicates like 43950/44534.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the strongest single representative issue overall: broad, specific, and already has close duplicate reports that point to the same regression.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Different failures: notebook custom-model init vs flash-attn2 fallback gating." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "Different models and symptoms; both are perplexity-related but not the same bug." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Qwen3Moe FP8 failure is unrelated to flash-attn2 fallback checks." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Different tied-weight issues in different model families; not the same concrete bug." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": true, + "reason": "Both describe the same remote-code/custom-module path collision class causing the wrong module to load." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization/offloading behavior is unrelated to tied-weights regressions." + }, + { + "left": "issue:35308", + "right": "issue:36331", + "accept": false, + "reason": "Different problems: iterable-dataset multi-GPU crash vs CustomTrainer compute_loss signature mismatch." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "Unrelated DSA/indexer discussion vs docs cleanup." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "A cache-strategy typo is unrelated to the removed CLI command doc issue." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Different model/trainer failures with no clear shared code path." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache crash is unrelated to tied-weight regressions." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Different tokenizer/runtime problems on different models." + }, + { + "left": "issue:45183", + "right": "issue:46463", + "accept": false, + "reason": "Both touch input validation, but the titles point to different concrete bugs and scopes." + }, + { + "left": "issue:35308", + "right": "issue:42915", + "accept": false, + "reason": "Multi-GPU dataset crash is unrelated to Qwen3Moe FP8 issues." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "Model-loading failure is unrelated to tied-weights regressions." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "Docs/example errors vs model quality regression are different issues." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Same non-persistent-buffer corruption regression, just reworded." + }, + { + "left": "issue:46489", + "right": "issue:46491", + "accept": false, + "reason": "Different tokenizer bugs affecting different models/symptoms." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Unexpected compute_loss kwarg vs ZeRO-3 loading crash are different bugs." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": true, + "reason": "Same regression: non-persistent buffers get corrupted/junk-filled after load/materialization." + }, + { + "left": "issue:33666", + "right": "issue:35308", + "accept": false, + "reason": "Multi-GPU training for Qwen2-VL is not the same as the iterable-dataset crash." + }, + { + "left": "issue:46032", + "right": "issue:47246", + "accept": true, + "reason": "Both describe the same Mamba2 inter-chunk recurrence bug, with NemotronH as a missing propagation of the same fix." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Different bitsandbytes/storage incompatibilities with different failure modes." + }, + { + "left": "issue:46899", + "right": "issue:46932", + "accept": true, + "reason": "Same Gemma4 quantized vision/audio embedder dtype-casting bug; 46932 is an explicit follow-up/dup of the same fix." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "Different tokenizer-related complaints; not the same concrete bug." + }, + { + "left": "issue:45920", + "right": "issue:46489", + "accept": false, + "reason": "General AutoTokenizer regression vs a specific DeepSeek-Coder tokenizer bug." + }, + { + "left": "issue:47332", + "right": "issue:47333", + "accept": true, + "reason": "Same offloading save failure for disk-offloaded Qwen3 models, just on different model variants." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Both mention SigLIP2, but one is zero probabilities and the other is degraded embeddings; not the same bug." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Torch autocast dtype mismatch and RLHF NaN from position_ids are different failure mechanisms." + }, + { + "left": "issue:45987", + "right": "issue:46858", + "accept": true, + "reason": "Same cache API contract issue: get_seq_length returns the wrong tensor form instead of an int." + }, + { + "left": "issue:43950", + "right": "issue:45902", + "accept": true, + "reason": "Same non-persistent-buffer initialization regression; 45902 is a model-specific manifestation." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analyst output is generally grounded in the packet and the soft-edge verdicts stay conservative. The accepted pairs mostly have clear title-level overlap in the same bug family (non-persistent buffers, cache length contract, Gemma4 dtype casting, offloading save failures, and related custom-module loading issues). I don\u2019t see a clear overstatement that would warrant rejection." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 5, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:24643", + "issue:28530", + "issue:41553", + "issue:41762", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43404", + "issue:43493", + "issue:43643", + "issue:43782", + "issue:43824", + "issue:43827", + "issue:43828", + "issue:43856", + "issue:43872", + "issue:43881", + "issue:43883", + "issue:43950", + "issue:43975", + "issue:44368", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44509", + "issue:44661", + "issue:44910", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45237", + "issue:45356", + "issue:45399", + "issue:45488", + "issue:45663", + "issue:45685", + "issue:45715", + "issue:45759", + "issue:45823", + "issue:45859", + "issue:45902", + "issue:46315", + "issue:46489", + "issue:46661", + "issue:46869" + ], + "soft_pairs": [ + "issue:43054|issue:43493", + "issue:44910|issue:45859", + "issue:45663|issue:45902", + "issue:44403|issue:44451", + "issue:42947|issue:44387", + "issue:24643|issue:28530", + "issue:43975|issue:45237", + "issue:41762|issue:42915", + "issue:44403|issue:44488", + "issue:43643|issue:45020", + "issue:44991|issue:45399", + "issue:42947|issue:43856", + "issue:45685|issue:45823", + "issue:45081|issue:45399", + "issue:43824|issue:43872", + "issue:45020|issue:45823", + "issue:44661|issue:45020", + "issue:43782|issue:44960", + "issue:41553|issue:43883", + "issue:43828|issue:45823", + "issue:45020|issue:45356", + "issue:43872|issue:46869", + "issue:42491|issue:44368", + "issue:41553|issue:43881", + "issue:45759|issue:46661", + "issue:45020|issue:46661", + "issue:45488|issue:46489", + "issue:43404|issue:43950", + "issue:45715|issue:46315", + "issue:43827|issue:44509", + "issue:43881|issue:43883" + ], + "prepared_review_unit_hash": "bc878288203857719e234ff00567e5a464f13f9d3e690c572e52f8c6663a3562", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37488, + "estimated_input_tokens": 9372, + "estimated_eval_tokens": 19000 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose collection of unrelated Transformers regressions around model loading, tokenizers, remote-code configs, MoE/LoRA training, and a couple of docs issues. Only the two v5 pipeline-doc tickets look like the same underlying change; the rest are too model- or code-path-specific to merge as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 45020 is the broadest representative: it covers the general class of recent v5 remote-code/model-loading breakages that several other issues touch, even though this cluster is not a single duplicate family.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the best overall issue to anchor the cluster because it is the clearest umbrella report, has strong discussion activity, and aligns with the dominant theme of v5 loading regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43054", + "right": "issue:43493", + "accept": false, + "reason": "Both are SigLIP2-related, but one is a quality/regression report and the other is an HF-vs-JAX implementation discrepancy; different bugs and fixes." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "One is a flash-attention crash from 3D position_ids handling; the other is CPU-offload incompatibility in rotary embedding. Related model family, different code paths." + }, + { + "left": "issue:45663", + "right": "issue:45902", + "accept": false, + "reason": "Gemma4 FSDP2 shared_kv_states bug vs Qwen3_5MoE inv_freq materialization bug; completely different models and failure modes." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "Generic load-noise complaint vs a specific model load failure for ScandiBERT; not the same underlying bug." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Gradient checkpointing with PEFT LoRA is a training-behavior issue; the other is int4 quantization causing higher reserved memory. Both memory-adjacent, but not the same bug." + }, + { + "left": "issue:24643", + "right": "issue:28530", + "accept": false, + "reason": "DeepSpeed weight-shape RuntimeError and early-stopping metric selection are unrelated training issues." + }, + { + "left": "issue:43975", + "right": "issue:45237", + "accept": false, + "reason": "Tokenizer detokenization regression vs AMD GPU support failure; no shared code path." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Gemma3 ZeRO-3 load failure vs Qwen3-MoE FineGrainedFP8Config failure; both loading-related, but distinct model/config bugs." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Both mention loading failures, but one is a generic transformer noise report and the other is a specific sleng-bert model load breakage." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "AutoConfig losing fields with trust_remote_code is a config-construction bug, not the broader remote-code loading regressions described in 45020." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading failure for est-roberta vs flash-attn fallback blocking; unrelated subsystems." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Both involve training memory/perf, but one is ineffective gradient checkpointing with LoRA and the other is inefficient Qwen3 MoE memory usage." + }, + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "MPS histogram op not implemented for Int vs Gemma4 device-mismatch under device_map=auto; different platform bugs." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Mistral regex patch crash in tokenizer loading vs flash-attn fallback gating; no common underlying change." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Missing Qwen2.5-VL import and bitsandbytes Int8Params incompatibility are unrelated breakages." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Remote-code model breakage and Gemma4 device-mapping mismatch are different issues." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Tokenizer-mapping-name failure in add-new-model-like is a registry bug, not the broader remote-code regression." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Qwen3VL weight_only load error and GLM5 are different model-specific failures." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Bad AutoTokenizer error message for Voxtral vs missing all_tied_weights_keys in Molmo; unrelated bugs." + }, + { + "left": "issue:43828", + "right": "issue:45823", + "accept": false, + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is a computation issue, while Gemma4 PLE has a device placement mismatch." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Remote-code breakage vs Kimi-K2.5 tokenizer codec regression; different symptoms and fixes." + }, + { + "left": "issue:43872", + "right": "issue:46869", + "accept": false, + "reason": "bitsandbytes parameter init incompatibility is unrelated to YOLO multi-GPU device-map failure." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "LoRA Qwen3-MoE incompatibility on hf4.x vs a tie_word_embeddings warning in ms-swift; same family, but not the same bug." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Voxtral tokenizer error-message issue vs GLM-4V load failure; different models and failure paths." + }, + { + "left": "issue:45759", + "right": "issue:46661", + "accept": false, + "reason": "AutoModel.from_config not unwrapping text_config vs remote-code models missing config_class; both model-construction issues, but not the same root cause." + }, + { + "left": "issue:45020", + "right": "issue:46661", + "accept": false, + "reason": "General remote-code regressions and a specific config_class AttributeError are related only at a very high level." + }, + { + "left": "issue:45488", + "right": "issue:46489", + "accept": false, + "reason": "DeepSeek tokenizer pre-tokenizer override vs DeepSeek-Coder v1 detokenization regression; related model family, different tokenizer bugs." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Mistral3 lm_head tying bug vs non-persistent buffer corruption in from_pretrained; unrelated code paths." + }, + { + "left": "issue:45715", + "right": "issue:46315", + "accept": false, + "reason": "Both are tokenizer performance regressions, but one is convert_ids_to_tokens and the other is convert_tokens_to_ids; not mergeable as one bug." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "These are the same docs regression: outdated pipeline-task docs still reference summarization/translation/text2text-generation after v5 removals." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "GLM-4V load failure and Molmo tied-weights attribute error are different model-specific issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the one accepted soft pair (43827/44509) looks mergeable as the same docs regression. The other soft-edge rejections are generally conservative. Minor wording note: 'strong discussion activity' for 45020 is a bit stronger than the packet evidence strictly shows, but it does not materially distort the cluster assessment." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 6, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36331", + "issue:41093", + "issue:41553", + "issue:42617", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43366", + "issue:43531", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43792", + "issue:43824", + "issue:43828", + "issue:43854", + "issue:43940", + "issue:43950", + "issue:43957", + "issue:44155", + "issue:44220", + "issue:44479", + "issue:44530", + "issue:44560", + "issue:44589", + "issue:44805", + "issue:44863", + "issue:44918", + "issue:44928", + "issue:44991", + "issue:45305", + "issue:45313", + "issue:45325", + "issue:45584", + "issue:45588", + "issue:45663", + "issue:45685", + "issue:45823", + "issue:45834", + "issue:45902", + "issue:45923", + "issue:46402", + "issue:46468", + "issue:46515", + "issue:46894", + "issue:46897" + ], + "soft_pairs": [ + "issue:43257|issue:45305", + "issue:43792|issue:44220", + "issue:36331|issue:41093", + "issue:46894|issue:46897", + "issue:42947|issue:45923", + "issue:43940|issue:45313", + "issue:24643|issue:30064", + "issue:44479|issue:44560", + "issue:43854|issue:44863", + "issue:41093|issue:44805", + "issue:43366|issue:43828", + "issue:43716|issue:46468", + "issue:44928|issue:45588", + "issue:43366|issue:43531", + "issue:44530|issue:44863", + "issue:43596|issue:46515", + "issue:44863|issue:46402", + "issue:45325|issue:45902", + "issue:43366|issue:44589", + "issue:43638|issue:46515", + "issue:43716|issue:44560", + "issue:45663|issue:45823", + "issue:41553|issue:44991", + "issue:42915|issue:43957", + "issue:44155|issue:45584", + "issue:43824|issue:43957", + "issue:42617|issue:43366", + "issue:44560|issue:45685", + "issue:43950|issue:45834", + "issue:44560|issue:44918" + ], + "prepared_review_unit_hash": "9133502e6ac13e37c9879c6ead242bc7481bcfb52f1ecae988d9d31d7496a5d6", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36767, + "estimated_input_tokens": 9192, + "estimated_eval_tokens": 18640 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is heterogeneous overall, but there are a few plausible duplicate pairs: Qwen ZeRO-3 weight-loading regressions, Qwen/VL video-input regressions, a repeated mask-shape IndexError, and a repeated training-loss double-shift bug. Most other pairs are clearly unrelated despite superficial subsystem overlap.", + "confidence": 0.66, + "canonical_issue_reason": "No single canonical issue fits the whole cluster because the items split into several unrelated bug groups. If one umbrella issue is needed for the strongest subgroup, issue 44479 is the best representative because it describes a concrete 5.3.0 video-input regression across multiple Qwen/VL variants.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44479 is the best global issue for its subgroup: it states the version-specific regression clearly, spans multiple affected models, and is more umbrella-like than the narrower model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Both mention DeepSpeed and Qwen, but one is a weight-conversion failure and the other is a GAS gradient-averaging bug; different code paths." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "Both are audio/Whisper-adjacent, but the titles point to different failure modes and there is no clear shared code-path bug from the available evidence." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "One is a Trainer API kwarg breakage; the other is a tensor mask/index shape mismatch. These are unrelated defects." + }, + { + "left": "issue:46894", + "right": "issue:46897", + "accept": true, + "reason": "Both describe the same training-loss double-shift pattern across different models, suggesting one shared loss-computation fix could cover both." + }, + { + "left": "issue:42947", + "right": "issue:45923", + "accept": false, + "reason": "Gradient-checkpointing ineffective with LoRA is not the same as a missing supports_gradient_checkpointing flag in a trust_remote_code model." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": true, + "reason": "Both are Qwen-family DeepSpeed ZeRO-3 weight-loading failures with missing parameters, so they plausibly share the same loading-path bug." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "DeepSpeed training weight shape error and void segmentation-map preprocessing are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": true, + "reason": "Both are Qwen/VL video-input regressions in transformers 5.3.0; 44560 looks like a specific manifestation of the broader video bug in 44479." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "GLM-4.7 unit-test loading failure and NemotronH checkpoint loading failure are separate model-loading issues." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "The titles describe the same mask-vs-indexed-tensor shape mismatch error; the numeric sizes differ but the underlying bug appears identical." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "GGUF gpt-oss support and a torch.autocast dtype mismatch on Phi-tiny-MoE are unrelated." + }, + { + "left": "issue:43716", + "right": "issue:46468", + "accept": false, + "reason": "An image-preprocessor dtype mismatch for Mistral-3 is not the same as a broken Nemotron example." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "One is a Qwen3.5 RLHF NaN/SDPA fallback issue; the other is a flash-attention AttributeError on sink-less models. Different failures and code paths." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "GGUF architecture support and Qwen3-MoE sliding-window behavior are unrelated." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash and NemotronH checkpoint loading are different bugs." + }, + { + "left": "issue:43596", + "right": "issue:46515", + "accept": false, + "reason": "Zero-3/BertModel out-of-bounds indexing and torch-version incompatibility are not the same defect." + }, + { + "left": "issue:44863", + "right": "issue:46402", + "accept": false, + "reason": "Both are load failures, but for different model families and different failure mechanisms." + }, + { + "left": "issue:45325", + "right": "issue:45902", + "accept": false, + "reason": "One is a Qwen2.5-VL rope-scaling bug; the other is uninitialized memory after meta/to_empty materialization. Different underlying problems." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "gpt-oss GGUF support and missing Float8 storage are unrelated compatibility issues." + }, + { + "left": "issue:43638", + "right": "issue:46515", + "accept": false, + "reason": "A ZeRO-3 init/indexing failure with Bert and a torch version incompatibility are different issues." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Image-preprocessor dtype mismatch and Qwen3-VL video StopIteration are both multimodal, but they are not the same bug." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "One is a per-layer shared_kv_states KeyError in training; the other is a device-mapping mismatch in forward. Not the same code-path problem." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "A bad tokenizer error message for Voxtral is not the same as an actual tokenizer load failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Qwen3Moe FP8 failure and meta-device loading breakage are separate model/runtime incompatibilities." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "AudioFlamingo3 track-leak inference corruption and Whisper empty-transcription failure after special-token alignment are different bugs." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Missing Qwen2_5_VL import and meta-device loading breakage are unrelated." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "3d_parallel.py runtime failure and GGUF gpt-oss support are unrelated." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Qwen video StopIteration and an MPS histogram Int limitation are different hardware/model bugs." + }, + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Non-persistent buffer corruption in from_pretrained and a long-OCR index error are unrelated." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Video-loading StopIteration and SFT trainer embedding unpacking failure are different failure modes." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The identified duplicate pairs are plausible from the titles/body signals, and the claim that the cluster is heterogeneous is well supported. The umbrella selection of issue 44479 is also reasonable because it is explicitly version-specific and multi-model, unlike the narrower reports." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 7, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:36010", + "issue:38175", + "issue:39692", + "issue:41553", + "issue:41720", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43317", + "issue:43550", + "issue:43596", + "issue:43632", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43720", + "issue:43994", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44560", + "issue:44661", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44910", + "issue:44960", + "issue:44977", + "issue:45030", + "issue:45276", + "issue:45313", + "issue:45357", + "issue:45507", + "issue:45672", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:46468", + "issue:46489", + "issue:46515", + "issue:46846", + "issue:46858", + "issue:46899", + "issue:47254" + ], + "soft_pairs": [ + "issue:44683|issue:46515", + "issue:36010|issue:42915", + "issue:43645|issue:43646", + "issue:44910|issue:46858", + "issue:41553|issue:44488", + "issue:43550|issue:44910", + "issue:45823|issue:46515", + "issue:42947|issue:47254", + "issue:44661|issue:45823", + "issue:45812|issue:46489", + "issue:44560|issue:44805", + "issue:44877|issue:45030", + "issue:38175|issue:43994", + "issue:43638|issue:44661", + "issue:43596|issue:44661", + "issue:44661|issue:44805", + "issue:42491|issue:43720", + "issue:45672|issue:46899", + "issue:41720|issue:42915", + "issue:43632|issue:45834", + "issue:43317|issue:44756", + "issue:42915|issue:44756", + "issue:38175|issue:39692", + "issue:43644|issue:44534", + "issue:42491|issue:45276", + "issue:44451|issue:44488", + "issue:44960|issue:44977", + "issue:45507|issue:46468", + "issue:45313|issue:45357", + "issue:44863|issue:46846" + ], + "prepared_review_unit_hash": "59aed0fd7beafe2b3d1fda86286ad02d94daf15546e17e295a23ccdf0424726d", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35964, + "estimated_input_tokens": 8991, + "estimated_eval_tokens": 18238 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are not duplicates; they share a subsystem but describe different failures. I only accepted a small set of clear same-bug matches, mostly exact or near-exact duplicates.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43644 is the clearest representative of the non-persistent-buffer regression, and 44534 is effectively the same report with a reworded title.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43644 is the strongest canonical issue here because it is precise, broadly phrased, and has an almost exact duplicate in 44534.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44683", + "right": "issue:46515", + "accept": false, + "reason": "Different problems: torch>=2.9 flex_attention compilation vs a torch 2.4.1 incompatibility." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "Unrelated failures: GenerationMixin import error vs Qwen3MoE FP8 config handling." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "Same underlying custom-model initialization regression in Transformers v5; notebook mention is just a usage variant." + }, + { + "left": "issue:44910", + "right": "issue:46858", + "accept": false, + "reason": "Different code paths: flash-attention 3D position_ids crash vs static-cache seq-length arithmetic on non-CUDA backends." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading complaints, but they concern different models and no shared concrete bug is evident." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different failures in different attention stacks and models." + }, + { + "left": "issue:45823", + "right": "issue:46515", + "accept": false, + "reason": "Device-map forward mismatch vs a general torch version incompatibility; not the same bug." + }, + { + "left": "issue:42947", + "right": "issue:47254", + "accept": false, + "reason": "Both involve PEFT and checkpointing, but one is ineffective checkpointing while the other is a ZeRO-3 checkpoint error." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "Tokenizer-mapping registration bug vs Gemma4 device placement issue." + }, + { + "left": "issue:45812", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and appear to be separate bugs." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Different symptom and path: video StopIteration vs tensor mask shape IndexError." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both are config-validation load failures, but for different models and likely different root causes." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 regression: both report incorrect/nonsensical outputs for google/siglip2-base-patch16-224." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Bert/DeepSpeed ZeRO-3 init error is unrelated to add-new-model-like tokenizer mapping." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "DeepSpeed ZeRO-3 Bert init IndexError is unrelated to tokenizer mapping registration." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Different bugs: tokenizer-mapping registration vs mask/tensor shape mismatch." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Different model/version compatibility issues with different failure modes." + }, + { + "left": "issue:45672", + "right": "issue:46899", + "accept": true, + "reason": "Same Gemma4 4-bit multimodal quantization bug; both point to unsafe dtype handling in embedders." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Different models and different runtime failures." + }, + { + "left": "issue:43632", + "right": "issue:45834", + "accept": false, + "reason": "Internal flag regression vs OCR input index error; no shared bug." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "Different resource/loading problems: dequantized gpu+cpu offload vs mmap OOM on Strix Halo." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3MoE FP8 failure is unrelated to mmap/OOM behavior." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Same model family, but one is bad inference output and the other is a documentation example with model/processor and quantization issues." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Near-exact duplicate: both report Transformers v5 filling non-persistent buffers with junk." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "LoRA/hf4.x vs hf5.x incompatibility is unrelated to Gemma4 embedding resize behavior." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both are loading failures, but for different checkpoints and without evidence of the same root cause." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Different model families and distinct attention/runtime issues." + }, + { + "left": "issue:45507", + "right": "issue:46468", + "accept": false, + "reason": "GraniteMoEHybrid invalid method call is unrelated to the Nemotron example failure." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Different Qwen3.5 bugs: ZeRO-3 weight loading vs incorrect save_pretrained visual encoder keys." + }, + { + "left": "issue:44863", + "right": "issue:46846", + "accept": false, + "reason": "NemotronH checkpoint loading and Qwen3.5 TP-plan issues are unrelated." + }, + { + "left": "issue:44910", + "right": "issue:44977", + "accept": false, + "reason": "Different models and different flash-attention/runtime symptoms." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 BertModel IndexError; the pretrained/non-pretrained detail does not change the underlying bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the accepted pairs are all reasonably conservative near-duplicates. I don\u2019t see an overstatement strong enough to reject this review." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 8, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42947", + "issue:43257", + "issue:43404", + "issue:43425", + "issue:43454", + "issue:43644", + "issue:43645", + "issue:43708", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44075", + "issue:44079", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44387", + "issue:44402", + "issue:44488", + "issue:44756", + "issue:44910", + "issue:44960", + "issue:44991", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45362", + "issue:45406", + "issue:45488", + "issue:45588", + "issue:45663", + "issue:45704", + "issue:45759", + "issue:45812", + "issue:45823", + "issue:45859", + "issue:46050", + "issue:46459", + "issue:46512", + "issue:46829", + "issue:46897", + "issue:46901" + ], + "soft_pairs": [ + "issue:46459|issue:46512", + "issue:39692|issue:43994", + "issue:44488|issue:44991", + "issue:43644|issue:43645", + "issue:43257|issue:43854", + "issue:43856|issue:43873", + "issue:42491|issue:45335", + "issue:42491|issue:45292", + "issue:43854|issue:43931", + "issue:44164|issue:44291", + "issue:43856|issue:44387", + "issue:44960|issue:46050", + "issue:45488|issue:45812", + "issue:43708|issue:44075", + "issue:45276|issue:45335", + "issue:43425|issue:43957", + "issue:44291|issue:44756", + "issue:45362|issue:45406", + "issue:45663|issue:45759", + "issue:41720|issue:44155", + "issue:39692|issue:44960", + "issue:44910|issue:45588", + "issue:44402|issue:44488", + "issue:43404|issue:43454", + "issue:46897|issue:46901", + "issue:43873|issue:44387", + "issue:42947|issue:45704", + "issue:45663|issue:45859", + "issue:44155|issue:45823", + "issue:43957|issue:44756", + "issue:44079|issue:46829" + ], + "prepared_review_unit_hash": "15b8449bdb0ffdf163e8c12d5162579de274389354afcda697057e7ea401bc69", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37280, + "estimated_input_tokens": 9320, + "estimated_eval_tokens": 18896 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is largely heterogeneous; none of the soft pairs are safe duplicate merges based on the visible titles alone. Most pairs share only a broad subsystem (loading, quantization, tokenization, training) but differ in model, code path, or failure mode.", + "confidence": 0.91, + "canonical_issue_reason": "If forced to pick one representative issue, issue:43404 is the clearest, most code-path-specific bug report in the set: it names the affected model, the concrete symptom (untied lm_head), and points to a likely fix location in weight tying logic.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43404 is the best standalone candidate because it is narrowly scoped, reproducible, and likely to produce a focused fix; the rest are either broader environment regressions or model-specific variations that are not duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46459", + "right": "issue:46512", + "accept": false, + "reason": "Both mention dtype-related model loading, but one is about ignored dtype during checkpoint loading and the other about stale config.dtype after from_config; different bugs and code paths." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Both involve SigLIP2, but one is a doc example with processor/model mismatch and quantization failure while the other is incorrect outputs from AutoModel/pipeline; not the same underlying defect." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading regressions, but they affect different models and likely different tokenizer files/code paths." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers 5.0 complaints, but one is about non-persistent buffers and the other about notebook-defined custom models; unrelated failures." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Both are model-loading issues, but they concern different models and different failure modes (Qwen3 MoE conversion vs GLM-4.7-Flash unit-test loading)." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "One is Qwen3 MoE training memory usage, the other is quantization/offloading behavior; broad memory theme only, not the same bug." + }, + { + "left": "issue:42491", + "right": "issue:45335", + "accept": false, + "reason": "Different models and different symptoms; both are embedding-related but not the same concrete bug." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "Same broad resize/token-embedding theme, but the reports target different model families and embedding surfaces." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Different models with different loading failures; no shared concrete code path is evident." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Both concern loading/save mechanics, but one is extra_state handling and the other is init_empty_weights/_is_hf_initialized argument handling; separate issues." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Both mention memory pressure, but one is training-memory inefficiency and the other is int4 quantization reserved-memory/OOM regression." + }, + { + "left": "issue:44960", + "right": "issue:46050", + "accept": false, + "reason": "Different model families and different persistence problems; no evidence of the same bug." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but one changes pre-tokenizer behavior for DeepSeek-family models and the other yields wrong IDs for Granite; distinct failures." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer resume/max_steps logic is unrelated to optimizer argument handling." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": false, + "reason": "Both are resize_token_embeddings complaints, but they affect different architectures and different embedding endpoints; too little evidence for one underlying bug." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch version incompatibility and meta-device loading failures are different regressions." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "init_empty_weights TypeError is unrelated to Strix Halo mmap OOM behavior." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Different models and different crash causes; no duplicate relationship is apparent." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "FSDP2 shared_kv_states bug and composite-config unwrapping bug are separate code paths." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different models and failure types; one is CUDA assert with device mapping, the other is batched inference leakage." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "No substantive overlap beyond both being model-specific issues." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Both involve flash attention, but one is a Qwen3.5 packed-sequence/position_ids crash and the other is an AttributeError on sink-less models; not the same bug." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "Different tokenizer/model-loading issues with no shared concrete root cause." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Both are lm_head tying problems, but they affect different model families and likely different initialization paths." + }, + { + "left": "issue:46897", + "right": "issue:46901", + "accept": false, + "reason": "Both describe a training-loss double-shift pattern, but they are for different models and no shared implementation bug is confirmed." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Both touch quantization/memory, but one is offloading behavior and the other is reserved-memory growth under int4 quantization; different regressions." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing/PEFT behavior is unrelated to a T5 RMSNorm memory leak." + }, + { + "left": "issue:45663", + "right": "issue:45859", + "accept": false, + "reason": "Different model families and different runtime issues (FSDP2 attention state vs CPU-offload incompatibility)." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "Both are multimodal/runtime issues, but one is track-level leakage in AudioFlamingo3 and the other is device mismatch in Gemma4 PLE; not duplicates." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device loading and mmap/OOM are separate platform regressions." + }, + { + "left": "issue:44079", + "right": "issue:46829", + "accept": false, + "reason": "ModelOutput key assignment and ImageClassifierOutput field mispopulation are different bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 9, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:42796", + "issue:43335", + "issue:43381", + "issue:43425", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43716", + "issue:43828", + "issue:43854", + "issue:43957", + "issue:44164", + "issue:44292", + "issue:44387", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44530", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44863", + "issue:44877", + "issue:44912", + "issue:44928", + "issue:45084", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45698", + "issue:45701", + "issue:45812", + "issue:45823", + "issue:46315", + "issue:46326", + "issue:46468", + "issue:46491", + "issue:46869", + "issue:46894", + "issue:46901", + "issue:47228", + "issue:47332" + ], + "soft_pairs": [ + "issue:44863|issue:46468", + "issue:43335|issue:45663", + "issue:44479|issue:44877", + "issue:44387|issue:44756", + "issue:45276|issue:45292", + "issue:36010|issue:39404", + "issue:44292|issue:44912", + "issue:44164|issue:44756", + "issue:43957|issue:47332", + "issue:43381|issue:44928", + "issue:45588|issue:45663", + "issue:43425|issue:47228", + "issue:43645|issue:44164", + "issue:42796|issue:46326", + "issue:44589|issue:44756", + "issue:43425|issue:43828", + "issue:45292|issue:45335", + "issue:43854|issue:45216", + "issue:43646|issue:43708", + "issue:44402|issue:44451", + "issue:45684|issue:45698", + "issue:43425|issue:46869", + "issue:43425|issue:43716", + "issue:44683|issue:45823", + "issue:44530|issue:45084", + "issue:46894|issue:46901", + "issue:45812|issue:46315", + "issue:46315|issue:46491", + "issue:39404|issue:45701" + ], + "prepared_review_unit_hash": "d856ebdf2ed388ab3799784b0e90df72012e2a87ba02dbf6b21291105a07c8e6", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36283, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous and does not form one clear duplicate set. I only see two plausible duplicate mini-clusters: the resize_token_embeddings regression family, and the training-loss double-shift family. Everything else looks like a separate bug or regression.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue for the entire set; the items split into unrelated bugs. If narrowing to the only clear duplicate family, issue 45292 is the broadest representative for the resize_token_embeddings regressions.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45292 is the cleanest issue to anchor the resize_token_embeddings subset because it is the most generic description and matches the model-specific variants.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44863", + "right": "issue:46468", + "accept": false, + "reason": "Both are Nemotron-related, but one is checkpoint loading and the other is an example breakage; not the same concrete bug." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "Different models and different failure modes; no shared code-path evidence." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Both are model-loading regressions, but one is Qwen-VL video input and the other is Granite Speech config strictness." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Both mention OOM, but the causes differ: int4 quantization memory growth vs Strix Halo mmap behavior." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same underlying resize_token_embeddings propagation bug; 45276 looks like a Gemma4-specific manifestation of the broader 45292 issue." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "ImportError for GenerationMixin and Whisper pipeline return_language are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Different quantization/model-loading regressions with different symptoms and code paths." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "save/from_pretrained extra_state handling is unrelated to Strix Halo mmap OOM." + }, + { + "left": "issue:43957", + "right": "issue:47332", + "accept": false, + "reason": "Meta-device model loading and disk-offloaded saving are different lifecycle paths." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Gradient checkpointing eval-mode restriction is unrelated to the RLHF NaN/SDPA fallback bug." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash-attention sink handling and Gemma4 FSDP2 shared_kv_states are separate issues." + }, + { + "left": "issue:43425", + "right": "issue:47228", + "accept": false, + "reason": "Torch version incompatibility and Sam3 bf16 compilation regression are not the same bug." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Custom-model notebook initialization and extra_state save/load failures are distinct." + }, + { + "left": "issue:42796", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer BPE-marker handling and mistral_common continue_final_message behavior are different issues." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to Strix Halo mmap OOM." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch 2.10 compatibility and Phi-tiny-MoE autocast dtype mismatch are unrelated." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Both describe the same resize_token_embeddings bug affecting downstream embedding modules in different model variants." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "GLM-4.7-Flash loading failure and Qwen3.5 save_pretrained regression are separate model-specific problems." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization breakage and Trainer max_steps miscalculation are unrelated." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Both involve model/tokenizer loading, but the titles point to different concrete failures and no shared bug is clear." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Read-only permission propagation and wrong custom-module loading after save_pretrained are different defects." + }, + { + "left": "issue:43425", + "right": "issue:46869", + "accept": false, + "reason": "Torch 2.10 incompatibility is unrelated to YOLO device_map auto multi-GPU failure." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Torch version compatibility and Mistral-3 image preprocessor dtype mismatch are different problems." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Compiled flex_attention regression and Gemma4 PLE device mismatch are unrelated." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash and non-template node compilation error are not the same code-path." + }, + { + "left": "issue:46894", + "right": "issue:46901", + "accept": true, + "reason": "Both are the same training-loss double-shift pattern: labels are shifted twice instead of using the unshifted labels." + }, + { + "left": "issue:45812", + "right": "issue:46315", + "accept": false, + "reason": "Wrong Granite token IDs and slow convert_tokens_to_ids are separate tokenizer regressions." + }, + { + "left": "issue:46315", + "right": "issue:46491", + "accept": false, + "reason": "Performance regression in token lookup is unrelated to CodeLlama spacing round-trip behavior." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper return_language pipeline breakage and tokenizer behavior changes are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: the cluster is mostly heterogeneous, and the only defensible duplicate families are the resize_token_embeddings trio (45276/45292/45335) and the training-loss double-shift pair (46894/46901). The soft-edge rejections are conservative, and the two acceptances are supported by closely aligned issue titles." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 10, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:36032", + "issue:39404", + "issue:40990", + "issue:42796", + "issue:42915", + "issue:43404", + "issue:43452", + "issue:43541", + "issue:43596", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43746", + "issue:43856", + "issue:43901", + "issue:43906", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44265", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44661", + "issue:44671", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44898", + "issue:44912", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45759", + "issue:45823", + "issue:46050", + "issue:46326", + "issue:46665", + "issue:46821", + "issue:46899", + "issue:47050", + "issue:47331", + "issue:47333" + ], + "soft_pairs": [ + "issue:43611|issue:43645", + "issue:44479|issue:44805", + "issue:44589|issue:45084", + "issue:44164|issue:45759", + "issue:43994|issue:44079", + "issue:40990|issue:47331", + "issue:44265|issue:44898", + "issue:36032|issue:43452", + "issue:39404|issue:46665", + "issue:43901|issue:44509", + "issue:44991|issue:45081", + "issue:46050|issue:46326", + "issue:44671|issue:44991", + "issue:43906|issue:46326", + "issue:43906|issue:44530", + "issue:39404|issue:44912", + "issue:45823|issue:46899", + "issue:44164|issue:46821", + "issue:43596|issue:44683", + "issue:44756|issue:47333", + "issue:43856|issue:44756", + "issue:44488|issue:44987", + "issue:36032|issue:42915", + "issue:43632|issue:47050", + "issue:44488|issue:45081", + "issue:43716|issue:45084", + "issue:43746|issue:47331", + "issue:42796|issue:43404", + "issue:43611|issue:43644", + "issue:44661|issue:44683", + "issue:43541|issue:45084" + ], + "prepared_review_unit_hash": "412b41472214e37c520b0833cb23bc1f6f475b68a297693892a09f8091cd68f2", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37279, + "estimated_input_tokens": 9320, + "estimated_eval_tokens": 18896 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: many issues are about distinct v5 regressions, model-specific loading bugs, or unrelated runtime errors. I did not find any soft pair that looks like the same underlying bug/change; the closest common theme is general Transformers v5 model-loading/initialization breakage.", + "confidence": 0.78, + "canonical_issue_reason": "issue:43611 is the broadest and most central representative of the only semi-coherent subtheme here: v5 model-loading/init regressions. Even so, the cluster is not a true duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43611 is the best single representative because it is general, well-scoped, and closest to the shared v5 loading/initialization regression theme seen in a few other issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both are v5 model-loading breakages, but one is base_model_prefix handling and the other is custom model initialization in notebooks; different failure paths." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Video-input regression in Qwen VL models vs a mask/tensor shape IndexError; unrelated code paths and symptoms." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Missing Float8 storage support vs a compiler/template-node TypeError; different subsystems and root causes." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "extra_state save/load handling vs unwrapping text_config in composite multimodal configs; different APIs and bugs." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "Model quality/regression report vs ModelOutput key assignment logic; not the same underlying issue." + }, + { + "left": "issue:40990", + "right": "issue:47331", + "accept": false, + "reason": "Perplexity on WikiText-2 vs MetalConfig batched quantized generation corruption; unrelated problems." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export failure around torch_compilable_check vs Perceiver non-default resolution interpolation; different model/export paths." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "Tokenizer method-name conflict vs gguf_file loading breakage; distinct tokenizer/model loading issues." + }, + { + "left": "issue:39404", + "right": "issue:46665", + "accept": false, + "reason": "Whisper pipeline return_language regression vs from_pretrained behavior inside Python classes; unrelated bugs." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs-related, but one is about return_all_scores and the other about removed pipeline tasks; different documentation mismatches." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer failures on different models: est-roberta loading vs Mistral regex patch crash." + }, + { + "left": "issue:46050", + "right": "issue:46326", + "accept": false, + "reason": "Quantized sinq save/load breakage vs continue_final_message on mistral_common tokenizers; different features and code paths." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT masked-LM prediction regression vs tokenizer loading failure for a different model; not the same bug." + }, + { + "left": "issue:43906", + "right": "issue:46326", + "accept": false, + "reason": "One is a reproduction of an older issue, the other is a Mistral-common tokenizer behavior bug; not duplicates." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "Isolated reproduction of 38071 vs PagedAttentionCache linear_attention crash; different failures." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline return_language regression vs MXFP4 quantization load fallback; unrelated functionality." + }, + { + "left": "issue:45823", + "right": "issue:46899", + "accept": false, + "reason": "Gemma4 device_map mismatch in forward vs unsafe dtype cast in multimodal embedders; different bugs." + }, + { + "left": "issue:44164", + "right": "issue:46821", + "accept": false, + "reason": "extra_state persistence bug vs TimesFM window_size AttributeError; unrelated APIs." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "DeepSpeed ZeRO3/BertModel init IndexError vs flex_attention compile failure on torch>=2.9; different stacks." + }, + { + "left": "issue:44756", + "right": "issue:47333", + "accept": false, + "reason": "Strix Halo mmap/OOM workaround vs saving a disk-offloaded Qwen3 model; different memory/save behaviors." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3 MoE training memory inefficiency vs Strix Halo mmap/OOM; same broad area, but not the same bug." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Two loading failures, but for different models and different root causes; not one concrete bug." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "Both are v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization; separate bugs." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Mistral-3 image preprocessing dtype mismatch vs a compile-time template-node error; unrelated." + }, + { + "left": "issue:43746", + "right": "issue:47331", + "accept": false, + "reason": "PEFT adapter checkpoint loading vs disk-offloaded quantized generation save/load; different features and symptoms." + }, + { + "left": "issue:42796", + "right": "issue:43404", + "accept": false, + "reason": "Tokenizer BPE marker handling vs lm_head weight tying in Mistral3 conditional generation; different code paths." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Model-registration/tokenizer-mapping bug vs flex_attention compile failure; unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Mixtral grouped_mm tracing error vs non-template-node compilation error; distinct compiler/runtime failures." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer load conflict vs Qwen3Moe FineGrainedFP8Config failure; unrelated." + }, + { + "left": "issue:43632", + "right": "issue:47050", + "accept": false, + "reason": "_is_hf_initialized flag regression vs ONNX export dtype mismatch in Canine attention; different problems." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Model loading failure for sleng-bert vs Mistral regex patch crash; different models and mechanisms." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 11, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42175", + "issue:42915", + "issue:43208", + "issue:43262", + "issue:43299", + "issue:43421", + "issue:43452", + "issue:43482", + "issue:43582", + "issue:43632", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43792", + "issue:43873", + "issue:43931", + "issue:44206", + "issue:44220", + "issue:44451", + "issue:44462", + "issue:44530", + "issue:44671", + "issue:44737", + "issue:44811", + "issue:44987", + "issue:44991", + "issue:45084", + "issue:45479", + "issue:45507", + "issue:45584", + "issue:45859", + "issue:46178", + "issue:46285", + "issue:46304", + "issue:46326", + "issue:46402", + "issue:46489", + "issue:46869", + "issue:46894", + "issue:46897", + "issue:47013", + "issue:47312", + "issue:47331" + ], + "soft_pairs": [ + "issue:45479|issue:46894", + "issue:44462|issue:46489", + "issue:46285|issue:46402", + "issue:43299|issue:43931", + "issue:42915|issue:44530", + "issue:44451|issue:44987", + "issue:45479|issue:46897", + "issue:43873|issue:47331", + "issue:36032|issue:43645", + "issue:44530|issue:46304", + "issue:44737|issue:45859", + "issue:40990|issue:43421", + "issue:40444|issue:46326", + "issue:34567|issue:40444", + "issue:42175|issue:46178", + "issue:43644|issue:43646", + "issue:45084|issue:45507", + "issue:43582|issue:43632", + "issue:44206|issue:44220", + "issue:44987|issue:44991", + "issue:39404|issue:44671", + "issue:43452|issue:43482", + "issue:43482|issue:47312", + "issue:44811|issue:45584", + "issue:43262|issue:47331", + "issue:43582|issue:47331", + "issue:43208|issue:47013", + "issue:42915|issue:45084", + "issue:43792|issue:46869" + ], + "prepared_review_unit_hash": "7c64ace01ee21771d6b863b48928221295bbbee503bc8e961a8b95c2eba779f5", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36314, + "estimated_input_tokens": 9079, + "estimated_eval_tokens": 18414 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous Transformers regressions with only loose topical overlap. None of the soft pairs look like safe duplicates under a strict triage standard, so I would not merge any of them. The closest subgroup is the v5/custom-model-initialization cluster, where issue 43646 is the best representative anchor.", + "confidence": 0.31, + "canonical_issue_reason": "issue:43646 is the broadest, most generic report in the only clearly related subgroup: a Transformers v5 regression breaking custom model initialization. It is the best umbrella-style issue among the cluster, even though the overall cluster is noisy and not a tight duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the most suitable canonical issue because it states the core user-facing failure in the most general terms and is likely to subsume narrower initialization-related reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45479", + "right": "issue:46894", + "accept": false, + "reason": "Both mention training-loss anomalies, but one is a zero-loss classification bug and the other is a sequence-shift bug in ASR training; different model classes and code paths." + }, + { + "left": "issue:44462", + "right": "issue:46489", + "accept": false, + "reason": "Both involve tokenizer behavior, but one is repository tokenizer loading and the other is a specific DeepSeek-Coder output regression after a prior fix; not the same bug." + }, + { + "left": "issue:46285", + "right": "issue:46402", + "accept": false, + "reason": "Different failures: CLIP text model structure inconsistency vs Sam3 text encoder weights not loading. No shared concrete code-path bug." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Both are Qwen-related loading issues, but one is a MoE loader breakage and the other is a weight-shape mismatch on a different VL model; too different to merge." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Qwen3Moe FP8 loading failure and Qwen3.5 paged-attention cache crash are distinct bugs in different subsystems." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Both are loading failures, but one is a model-specific load problem and the other is a transformers>=5.1 incompatibility for a different model; not the same underlying defect." + }, + { + "left": "issue:45479", + "right": "issue:46897", + "accept": false, + "reason": "Same general training-loss theme, but these are different model families and different shifting behavior; not one concrete bug." + }, + { + "left": "issue:43873", + "right": "issue:47331", + "accept": false, + "reason": "Offloading with quantization is a broad runtime issue, while the other is a specific MetalConfig batched generate corruption bug; not the same fix." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer special-method conflict and notebook custom-model initialization breakage are unrelated failures." + }, + { + "left": "issue:44530", + "right": "issue:46304", + "accept": false, + "reason": "PagedAttentionCache/Qwen3.5 crash vs distributed-build support regression; different code paths and symptoms." + }, + { + "left": "issue:44737", + "right": "issue:45859", + "accept": false, + "reason": "XLNet CPU device placement bug and Qwen3.5 CPU offload incompatibility are unrelated model/runtime issues." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "High perplexity on a model and runtime special-token post-processor updates are not the same underlying bug." + }, + { + "left": "issue:40444", + "right": "issue:46326", + "accept": false, + "reason": "Qwen2.5-VL multi-image iterable-dataset failure and mistral_common continue_final_message handling are distinct problems." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "TrainerState token accounting and Qwen2.5-VL multi-image finetuning are unrelated issues." + }, + { + "left": "issue:42175", + "right": "issue:46178", + "accept": false, + "reason": "Backend packaging/install problem versus BACKENDS_MAPPING validation error; different failure modes." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Both are Transformers v5 regressions around custom model/init behavior, but one is non-persistent buffer corruption and the other is custom model initialization failure; too different to treat as one bug." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template compilation TypeError and GraniteMoEHybrid invalid-method call are unrelated implementation bugs." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Apple Silicon allocator TypeError and _is_hf_initialized flag breakage are different regressions." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Both are audio-feature related, but one is a regression in LaserFeatureExtractor center handling and the other is a different fbank extraction issue; not clearly the same bug." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Both are transformers>=5 loading issues, but they target different models and different tokenizer/loading paths." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline return_language regression and CamemBERT masked-LM prediction regression are unrelated model behaviors." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": false, + "reason": "GGUF loading breaks in a generic from_pretrained path vs a specific Qwen2.5-GGUF v5 failure; related area, but not the same concrete bug." + }, + { + "left": "issue:43482", + "right": "issue:47312", + "accept": false, + "reason": "A Qwen2.5-GGUF loading regression and a kernels 0.16 incompatibility do not appear to share the same code-path problem." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Whisper batch_decode skip_special_tokens handling and empty-transcription generation after align_special_tokens are different bugs, though both are in Whisper." + }, + { + "left": "issue:43262", + "right": "issue:47331", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and MetalConfig quantization batch corruption are unrelated." + }, + { + "left": "issue:43582", + "right": "issue:47331", + "accept": false, + "reason": "Apple Silicon warmup TypeError and MetalConfig batched generate corruption are separate backend-specific issues." + }, + { + "left": "issue:43208", + "right": "issue:47013", + "accept": false, + "reason": "Both are xLSTM issues, but one is a collection of training blockers and the other is a specific return_last_states=False unpacking crash; not one merged fix." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3Moe FP8 loading failure and template compilation TypeError are unrelated." + }, + { + "left": "issue:43792", + "right": "issue:46869", + "accept": false, + "reason": "Whisper-large-v2 run failure and YOLO multi-GPU device-map auto failure are different models and different runtime paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 12, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28530", + "issue:34567", + "issue:36032", + "issue:40444", + "issue:41720", + "issue:42915", + "issue:43089", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43425", + "issue:43482", + "issue:43531", + "issue:43540", + "issue:43632", + "issue:43645", + "issue:43646", + "issue:43742", + "issue:43746", + "issue:43819", + "issue:43994", + "issue:44008", + "issue:44062", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44464", + "issue:44466", + "issue:44479", + "issue:44534", + "issue:44589", + "issue:44610", + "issue:44849", + "issue:44877", + "issue:44987", + "issue:45081", + "issue:45325", + "issue:45923", + "issue:45987", + "issue:46326", + "issue:46661", + "issue:46821", + "issue:46899", + "issue:47254" + ], + "soft_pairs": [ + "issue:44403|issue:45081", + "issue:43746|issue:46326", + "issue:43645|issue:46661", + "issue:43317|issue:43482", + "issue:42915|issue:44291", + "issue:44987|issue:45081", + "issue:43646|issue:44534", + "issue:43262|issue:43377", + "issue:36032|issue:43742", + "issue:43377|issue:43819", + "issue:42915|issue:44589", + "issue:43994|issue:44610", + "issue:43531|issue:44877", + "issue:28530|issue:34567", + "issue:44451|issue:45081", + "issue:44062|issue:44987", + "issue:43404|issue:44466", + "issue:36032|issue:41720", + "issue:44265|issue:44610", + "issue:44008|issue:46899", + "issue:44849|issue:45325", + "issue:44479|issue:45325", + "issue:43632|issue:46821", + "issue:43425|issue:44403", + "issue:43322|issue:43540", + "issue:43089|issue:45987", + "issue:43296|issue:43531", + "issue:45923|issue:47254", + "issue:44464|issue:45987", + "issue:40444|issue:45325" + ], + "prepared_review_unit_hash": "06eead36d32b8044a8fba94c313cd3ae39b396d60d84a75e3ef6de8191e3cecc", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36943, + "estimated_input_tokens": 9236, + "estimated_eval_tokens": 18728 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of unrelated Transformers bugs, with only loose thematic overlap around v5/init/load regressions. No soft pair looks like the same underlying bug strongly enough to merge, so all soft edges are rejected.", + "confidence": 0.68, + "canonical_issue_reason": "Issue 43632 is the closest thing to a broad umbrella bug here: it describes a concrete v5 initialization regression, which is the most reusable anchor among several nearby init/load-related reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43632 is the best single issue anchor because it is concrete, broadly phrased, and closer to a root-cause regression than the more model-specific or symptom-only reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Generic loading noise vs a tokenizer regex patch crash; different failure modes and code paths." + }, + { + "left": "issue:43746", + "right": "issue:46326", + "accept": false, + "reason": "PEFT local checkpoint loading and mistral_common continue_final_message are unrelated bugs." + }, + { + "left": "issue:43645", + "right": "issue:46661", + "accept": false, + "reason": "Both involve initialization, but one is notebook custom-model setup and the other is remote-code config_class resolution." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "Dequantized model offload failure and Qwen2.5-GGUF loading failure are distinct model-loading problems." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "Both touch v5/weights setup, but one is FineGrainedFP8Config on Qwen3Moe and the other is an init_empty_weights argument mismatch." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "A model load regression and a tokenizer regex crash are not the same underlying bug." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Custom model initialization and non-persistent buffer corruption are different issues." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio chat-template sampling-rate handling and MIMI padding-mask batching are different modality bugs." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "T5Tokenizer special-token method conflict and MobileLLM key error are unrelated." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "MIMI padding-mask support and DAC from_latents STE mismatch are separate model-specific defects." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "Float8 config failure and missing Float8 storage registration are only loosely related by dtype." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "Siglip2 bad outputs and OmDet-Turbo processor size mismatch are different model/processor bugs." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and granite_speech config strictness are unrelated." + }, + { + "left": "issue:28530", + "right": "issue:34567", + "accept": false, + "reason": "Early stopping metric selection and num_input_tokens_seen updates are different trainer bookkeeping issues." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "A model loading failure and a tokenizer patch crash are not the same bug." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "AddedToken keyword duplication and a model load regression have no concrete overlap." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "Tied lm_head weights at model definition time versus serialization inconsistency are related in theme but not the same code-path bug." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Tokenizer special-token conflict and Qwen3 auto device mapping cuda assert are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export/torch_compilable_check failure and OmDet input-size mismatch are separate problems." + }, + { + "left": "issue:44008", + "right": "issue:46899", + "accept": false, + "reason": "Both are multimodal, but one is a variable-name AttributeError and the other is a dtype-cast issue in embedders." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen variants and different bugs: hidden-states output vs rope index scaling." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Video-input regression across Qwen models and still-image position-id scaling are not the same defect." + }, + { + "left": "issue:43632", + "right": "issue:46821", + "accept": false, + "reason": "_is_hf_initialized initialization break and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch version incompatibility and noisy loading output are completely different issues." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Both are multimodal loading bugs, but they involve different models and different concrete failures." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Generation performance overhead and StaticCache return-type mismatch are unrelated." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and Qwen3-MoE sliding_window behavior are different bugs." + }, + { + "left": "issue:45923", + "right": "issue:47254", + "accept": false, + "reason": "Missing supports_gradient_checkpointing and a PEFT/DeepSpeed checkpoint error are both about checkpointing, but not the same issue." + }, + { + "left": "issue:44464", + "right": "issue:45987", + "accept": false, + "reason": "Chunked-generation inconsistency with compiled forward and StaticCache.get_seq_length type mismatch are different code paths." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "IterableDataset multi-image finetuning failure and Qwen2.5-VL temporal position scaling are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 13, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:43296", + "issue:43381", + "issue:43421", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43611", + "issue:43632", + "issue:43716", + "issue:43746", + "issue:43792", + "issue:43872", + "issue:43883", + "issue:44079", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44387", + "issue:44403", + "issue:44488", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44991", + "issue:45072", + "issue:45081", + "issue:45216", + "issue:45357", + "issue:45542", + "issue:45663", + "issue:45812", + "issue:45923", + "issue:46315", + "issue:46491", + "issue:46710", + "issue:46821", + "issue:46899", + "issue:47404" + ], + "soft_pairs": [ + "issue:44220|issue:44683", + "issue:43872|issue:44291", + "issue:43716|issue:46899", + "issue:43611|issue:44534", + "issue:44403|issue:44991", + "issue:44991|issue:46315", + "issue:44291|issue:44589", + "issue:45216|issue:45357", + "issue:43883|issue:44488", + "issue:43792|issue:44589", + "issue:45812|issue:46491", + "issue:42796|issue:43746", + "issue:33666|issue:40444", + "issue:43381|issue:45923", + "issue:43632|issue:44756", + "issue:40990|issue:44811", + "issue:43531|issue:44291", + "issue:35707|issue:40444", + "issue:44079|issue:45663", + "issue:42796|issue:46710", + "issue:41720|issue:45081", + "issue:43482|issue:43525", + "issue:43296|issue:46821", + "issue:42175|issue:43296", + "issue:44387|issue:47404", + "issue:43883|issue:44991", + "issue:42175|issue:45542", + "issue:43421|issue:44568", + "issue:44079|issue:44534", + "issue:44265|issue:45072" + ], + "prepared_review_unit_hash": "490733add68c62e0c34f53e1182a38e59822bef0708c681d3e37dc49114233af", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36926, + "estimated_input_tokens": 9232, + "estimated_eval_tokens": 18720 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose similarity set of unrelated issues. It does contain two accepted duplicate/follow-up pairs: the `_is_hf_initialized` compatibility regression and the Qwen3.5 `save_pretrained` visual-key regression. The rest are different models, subsystems, or failure modes.", + "confidence": 0.88, + "canonical_issue_reason": "issue:44291 is the clearest, most general statement of the `_is_hf_initialized` loading regression and best captures the root cause for that duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44291 is the strongest representative issue for the `_is_hf_initialized` regression in this cluster; it is the best anchor for that pair, though the cluster also includes a separate accepted Qwen3.5 `save_pretrained` pair.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Different bugs: one is about `_torch_extract_fbank_features()`, the other about `torch.export`/`flex_attention` on torch>=2.9." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Same underlying compatibility break around `_is_hf_initialized`; one report is bitsandbytes-specific, the other is the generic upstream regression." + }, + { + "left": "issue:43716", + "right": "issue:46899", + "accept": false, + "reason": "Both are multimodal dtype problems, but they affect different models and code paths; not the same concrete bug." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "Unrelated regressions: base_model_prefix loading vs non-persistent buffer initialization." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Different failure modes: generic load noise vs tokenizer loading failure for a specific model." + }, + { + "left": "issue:44991", + "right": "issue:46315", + "accept": false, + "reason": "Tokenizer loading failure is not the same as slower `convert_tokens_to_ids`; different code paths." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "`_is_hf_initialized` constructor incompatibility is unrelated to the float8 storage lookup error." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Same Qwen3.5 `save_pretrained` regression; the second is a narrower follow-up about the incorrect visual encoder keys." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing `all_tied_weights_keys` on Molmo is unrelated to loading `cjvt/sleng-bert`." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper runtime failure and float8 storage lookup error are different bugs." + }, + { + "left": "issue:45812", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." + }, + { + "left": "issue:42796", + "right": "issue:43746", + "accept": false, + "reason": "Tokenizer BPE-marker handling is unrelated to GraniteSpeech PEFT local-checkpoint loading." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both are Qwen-VL training issues, but one is multi-GPU training and the other is iterable-dataset multi-image batching." + }, + { + "left": "issue:43381", + "right": "issue:45923", + "accept": false, + "reason": "Eval-mode gradient-checkpointing behavior is not the same bug as a missing `supports_gradient_checkpointing` flag." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "Core v5 `_is_hf_initialized` breakage is unrelated to mmap OOM behavior on Strix Halo." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "Perplexity regression and Whisper `batch_decode` special-token handling are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior is unrelated to the `_is_hf_initialized` loading regression." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is a different bug from Qwen2.5-VL iterable-dataset failures." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment and Gemma4 FSDP2 shared-KV rebuild are different defects." + }, + { + "left": "issue:42796", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer BPE-marker handling is unrelated to the DeepSeek accuracy regression." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Auto device-map CUDA assert and Mistral regex patch crash are different code paths." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "GGUF loading failure and missing `pad_token_id` on Llama4Config are unrelated." + }, + { + "left": "issue:43296", + "right": "issue:46821", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and TimesFM `window_size` attribute errors are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "TensorFlow backend packaging and PaddleOCR-VL/vLLM loading are different issues." + }, + { + "left": "issue:44387", + "right": "issue:47404", + "accept": false, + "reason": "Both mention OOM, but one is int4 reserved-memory regression and the other is device_map conversion transients." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Molmo tied-weights attribute error is unrelated to tokenizer loading for EMBEDDIA/est-roberta." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both touch TensorFlow availability, but one is missing TF in the torch extra and the other is a different backend-init problem when only tensorboard is present." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime special-token post-processor syncing is not the same as add_special_tokens failing for one tokenizer." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer corruption are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "torch.export + torch_compilable_check is unrelated to bfloat16 dtype mismatches in inference." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary is mostly grounded, and the soft-edge rejections are conservative. However, the rationale for issue:44291 overstates the structure of the cluster: it says 44291 is the anchor for the \"only clear duplicate pair,\" but the packet also supports a second accepted duplicate/follow-up pair for Qwen3.5 save_pretrained (45216/45357). Please revise that wording so it only claims 44291 is the best anchor for the _is_hf_initialized pair, not the only duplicate pair in the cluster." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 14, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42796", + "issue:43262", + "issue:43317", + "issue:43377", + "issue:43454", + "issue:43482", + "issue:43606", + "issue:43611", + "issue:43644", + "issue:43646", + "issue:43746", + "issue:43818", + "issue:43828", + "issue:43856", + "issue:43873", + "issue:43994", + "issue:44079", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44462", + "issue:44466", + "issue:44479", + "issue:44509", + "issue:44829", + "issue:44877", + "issue:44898", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45491", + "issue:45538", + "issue:45584", + "issue:45701", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:46402", + "issue:46697", + "issue:46821", + "issue:46909", + "issue:47217" + ], + "soft_pairs": [ + "issue:43482|issue:45538", + "issue:44462|issue:45812", + "issue:43611|issue:43646", + "issue:45325|issue:45834", + "issue:43317|issue:43856", + "issue:44291|issue:44877", + "issue:44509|issue:46909", + "issue:40444|issue:45381", + "issue:43644|issue:44079", + "issue:43377|issue:45491", + "issue:45325|issue:45381", + "issue:43994|issue:44898", + "issue:40444|issue:47217", + "issue:43454|issue:44466", + "issue:43873|issue:45701", + "issue:45325|issue:46697", + "issue:43606|issue:44265", + "issue:43818|issue:46402", + "issue:43818|issue:45357", + "issue:44479|issue:45823", + "issue:42796|issue:45381", + "issue:43828|issue:46821", + "issue:40990|issue:45584", + "issue:44829|issue:45479", + "issue:43262|issue:43746", + "issue:41720|issue:45823", + "issue:44186|issue:44265", + "issue:35707|issue:41720", + "issue:34634|issue:35707" + ], + "prepared_review_unit_hash": "892b22bc5e289e95ee78f993df9ff3cac033263f8d6ca02c32d28a3d4fcbe935", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36675, + "estimated_input_tokens": 9169, + "estimated_eval_tokens": 18594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is only loosely related: most items are separate Transformers v5 regressions across tokenization, model loading, multimodal inputs, device mapping, and docs. None of the soft edges look like true duplicates of the same underlying bug or change.", + "confidence": 0.23, + "canonical_issue_reason": "Issue 43611 is the broadest representative of the cluster\u2019s recurring v5 model-loading breakages; several other items are adjacent to loading/init/serialization regressions, but the cluster is not a single duplicate bug.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43611 is the best stand-in for the cluster because it is the most generic and central of the v5 loading regressions, though it does not actually subsume most of the other issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Both are tokenizer/loading-related, but one is a GGUF loading failure and the other is a CLIPTokenizer max-length bug; different code paths and symptoms." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": false, + "reason": "Both involve AutoTokenizer behavior, but one ignores tokenizer.json while the other returns wrong token IDs for Granite models; not the same bug." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are v5 loading regressions, but one is about base_model_prefix loading and the other about custom model initialization; distinct failures." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Both are multimodal indexing bugs, but one is Qwen2.5-VL rope scaling for still images and the other is Kosmos2.5 OCR index overflow; different issues." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "One is a device-map/offload loading failure, the other is Qwen3 MoE training memory inefficiency; not the same underlying bug." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "Both affect loading/config initialization, but one is an init_empty_weights argument error and the other is strict config rejection for granite_speech; separate root causes." + }, + { + "left": "issue:44509", + "right": "issue:46909", + "accept": false, + "reason": "Both are documentation-related, but they cover unrelated docs problems (removed pipeline tasks vs a broken DIA example)." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL multimodal bugs, but one is finetuning with multiple images and the other is video vision_position_ids; different paths." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "One concerns non-persistent buffer serialization, the other ModelOutput key assignment; unrelated internals." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Both involve batching/padding side effects, but they are different models and different failures: missing padding-mask support vs NaN embeddings from sliding-window attention." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL positional-id issues, but one is still-image temporal scaling and the other is video input vision_position_ids; not the same concrete bug." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Different models and bugs: SigLIP2 nonsense outputs with AutoModel/pipeline versus Perceiver non-default-resolution failure." + }, + { + "left": "issue:40444", + "right": "issue:47217", + "accept": false, + "reason": "Both mention image placement in VL prompts, but one is finetuning with multiple images per prompt and the other is user-text placeholder handling; not clearly the same bug." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Both touch lm_head weights, but one is missing weight tying in a config and the other is inconsistent save_pretrained serialization depending on device." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "One is a quantization/offloading runtime issue and the other is a tokenizer behavior change; unrelated." + }, + { + "left": "issue:45325", + "right": "issue:46697", + "accept": false, + "reason": "Different subsystems entirely: Qwen2.5-VL rope indexing versus Swin2SR image preprocessing padding." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "Bark CPU-offload device mismatch and torch.export failing on torch_compilable_check are unrelated problems." + }, + { + "left": "issue:43818", + "right": "issue:46402", + "accept": false, + "reason": "Different models and failures: Video-LLaVA temporal-attention/weight-sharing bug versus Sam3Model text encoder weights not loading." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Video-LLaVA architecture bug and Qwen3.5 save_pretrained key regression are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Both are device/runtime regressions, but one is a Qwen video-input regression and the other is a Gemma4 PLE device mismatch during forward." + }, + { + "left": "issue:42796", + "right": "issue:45381", + "accept": false, + "reason": "Tokenizer BPE marker handling and Qwen2.5-VL video vision_position_ids are unrelated bugs." + }, + { + "left": "issue:43828", + "right": "issue:46821", + "accept": false, + "reason": "Different failure modes in different models: autocast dtype mismatch versus a missing window_size attribute." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "High perplexity on WikiText-2 and Whisper empty-transcription failure are unrelated." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both are degenerate training/loss complaints, but one is flash_attention_3 behavior and the other is single-label classification with num_labels=1; different code paths." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio processor chat-template sample-rate defaulting and GraniteSpeech PEFT checkpoint loading are separate issues." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Both involve device_map/forward placement, but they affect different models and distinct errors." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer crashes on NER/padding and torch.export failures on torch_compilable_check are unrelated." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device-mapping cuda asserts." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "Bark voice_preset not working and progressive generation inputs_embeds/past_key_values are different bugs with no shared underlying code path." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 15, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:36331", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42617", + "issue:42796", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43450", + "issue:43454", + "issue:43550", + "issue:43572", + "issue:43582", + "issue:43606", + "issue:43720", + "issue:43749", + "issue:43819", + "issue:43873", + "issue:44186", + "issue:44387", + "issue:44534", + "issue:44568", + "issue:44610", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45325", + "issue:45685", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45823", + "issue:45920", + "issue:45923", + "issue:46050", + "issue:46209", + "issue:46315", + "issue:46821", + "issue:47404", + "issue:47405" + ], + "soft_pairs": [ + "issue:43606|issue:43819", + "issue:42617|issue:42915", + "issue:45823|issue:46821", + "issue:43454|issue:43550", + "issue:40990|issue:42947", + "issue:43257|issue:45923", + "issue:43454|issue:45072", + "issue:47404|issue:47405", + "issue:44568|issue:45920", + "issue:43377|issue:44610", + "issue:36331|issue:42617", + "issue:42796|issue:46209", + "issue:44877|issue:44991", + "issue:36331|issue:42915", + "issue:43454|issue:43720", + "issue:41720|issue:43322", + "issue:43749|issue:44534", + "issue:41720|issue:42617", + "issue:43377|issue:44898", + "issue:43317|issue:43873", + "issue:41720|issue:44977", + "issue:43572|issue:45759", + "issue:45325|issue:46315", + "issue:44387|issue:45704", + "issue:40990|issue:45701", + "issue:43322|issue:46050", + "issue:43450|issue:45701", + "issue:41553|issue:43582", + "issue:45685|issue:46821", + "issue:43873|issue:44756", + "issue:44186|issue:45072" + ], + "prepared_review_unit_hash": "019053931ebc3676e4dc36aa41f510a0e70971ec3bbef35c35c49b008708d49a", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36435, + "estimated_input_tokens": 9109, + "estimated_eval_tokens": 18474 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous cluster of mostly unrelated issues across tokenizer regressions, model-loading/offloading, quantization, and model-specific runtime bugs. I do not see any soft pair that is clearly the same underlying defect, so all soft links should be rejected. If a representative issue is needed, 43873 is the broadest offloading/quantization report, but it does not explain the whole cluster.", + "confidence": 0.9, + "canonical_issue_reason": "43873 is the broadest representative of the offloading/quantization subset and is the most central active loading-related report, but the cluster overall is too mixed to have a perfect canonical issue.", + "canonical_pr_reason": null, + "best_issue_reason": "43873 is the most generally useful issue in the set: open, broad, and closest to the repeated offloading/device_map/quantization failures seen here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "Different failures: CPU-offload device mismatch for Bark vs DAC latent/STE mismatch in an audio codec path." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "One is a 3d_parallel.py runtime problem; the other is a Qwen3Moe FineGrainedFP8Config failure." + }, + { + "left": "issue:45823", + "right": "issue:46821", + "accept": false, + "reason": "Gemma4 PLE device mismatch is unrelated to TimesFM's missing window_size attribute." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "AyaVision lm_head weight tying and Bamba torch.compile+SDPA are separate code paths." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "GPT-OSS perplexity regression is not the same bug as LoRA gradient-checkpointing ineffectiveness." + }, + { + "left": "issue:43257", + "right": "issue:45923", + "accept": false, + "reason": "Accelerate/deepspeed MOE weight conversion differs from a missing gradient_checkpointing flag." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "AyaVision weight tying is unrelated to bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:47404", + "right": "issue:47405", + "accept": false, + "reason": "Quantized loading OOM from destination-GPU transients is distinct from silent wrong-Auto-class conversion skip." + }, + { + "left": "issue:44568", + "right": "issue:45920", + "accept": false, + "reason": "Specific add_special_tokens BOS/EOS regression is different from broader wrong-token-id regressions." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "MIMI padding-mask batching bug is unrelated to OmDet-Turbo's image-size mismatch." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Trainer compute_loss signature mismatch and 3d_parallel.py failure are unrelated." + }, + { + "left": "issue:42796", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are different bugs." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config rejection for granite_speech and tokenizer loading failure for est-roberta are different load paths." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "Trainer API keyword mismatch and Qwen3Moe FP8 failure are unrelated." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "AyaVision lm_head tying bug is not the same as BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Qwen3 auto device-map cuda assert and Llava Next segfault are different model/load failures." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are separate regressions." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Qwen3 auto device mapping assert is unrelated to 3d_parallel.py failing to run." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "MIMI batched padding-mask bug and Perceiver non-default resolution failure are unrelated." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Both involve offloading/quantization, but one is a dequantized-model load failure and the other is a broader offloading complaint; not clearly the same bug." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Auto device-map crash and flash-attention generation issue in Qwen3.5 are different symptoms/code paths." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "StableLm pad_token_idx regression and Qwen multimodal text_config unwrapping are unrelated config bugs." + }, + { + "left": "issue:45325", + "right": "issue:46315", + "accept": false, + "reason": "Qwen2.5-VL rope-index scaling and slow convert_tokens_to_ids are different tokenizer/model behaviors." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "int4 quantization OOM from reserved memory is not the same as T5's apex RMSNorm memory leak." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "GPT-OSS perplexity regression and tokenizer-version tokenization changes are different issues." + }, + { + "left": "issue:43322", + "right": "issue:46050", + "accept": false, + "reason": "Llava Next loading segfault and sinq save/load breakage are unrelated quantized-model bugs." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Video processor batch-shape bug and tokenizer-version tokenization change do not match." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Bad AutoTokenizer error message for Voxtral is unrelated to Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:45685", + "right": "issue:46821", + "accept": false, + "reason": "MPS histogram on Int and TimesFM window_size attribute error are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Strix Halo mmap OOM is a hardware-specific mmap issue, not the same as quantization offloading regression." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "LayoutLMv2 tokenizer crash and bfloat16 dtype mismatch are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 16, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:26413", + "issue:28530", + "issue:30064", + "issue:33666", + "issue:41720", + "issue:42915", + "issue:43065", + "issue:43295", + "issue:43317", + "issue:43404", + "issue:43441", + "issue:43531", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43646", + "issue:43708", + "issue:43716", + "issue:43720", + "issue:43881", + "issue:44008", + "issue:44186", + "issue:44361", + "issue:44423", + "issue:44451", + "issue:44466", + "issue:44488", + "issue:44610", + "issue:44898", + "issue:45161", + "issue:45357", + "issue:45406", + "issue:45701", + "issue:45704", + "issue:45823", + "issue:46291", + "issue:46402", + "issue:46612", + "issue:46665", + "issue:46821", + "issue:46858", + "issue:46869", + "issue:46897", + "issue:47020", + "issue:47317", + "issue:47337", + "issue:47404" + ], + "soft_pairs": [ + "issue:44466|issue:45357", + "issue:43295|issue:45701", + "issue:26413|issue:28530", + "issue:44451|issue:47020", + "issue:43065|issue:46402", + "issue:43550|issue:43720", + "issue:43550|issue:43606", + "issue:43577|issue:43708", + "issue:46821|issue:46869", + "issue:44186|issue:44610", + "issue:43646|issue:46665", + "issue:46612|issue:46858", + "issue:41720|issue:45161", + "issue:46897|issue:47317", + "issue:44008|issue:45823", + "issue:42915|issue:43317", + "issue:44423|issue:45406", + "issue:44610|issue:44898", + "issue:28530|issue:33666", + "issue:42915|issue:46291", + "issue:43404|issue:45357", + "issue:43531|issue:43572", + "issue:45704|issue:47404", + "issue:43716|issue:46821", + "issue:28530|issue:30064", + "issue:43881|issue:44488", + "issue:43441|issue:47337", + "issue:44186|issue:44361" + ], + "prepared_review_unit_hash": "3f127c80bc7bc9728727b0766640ba74556052e6665316e04a6a56b976f96270", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36200, + "estimated_input_tokens": 9050, + "estimated_eval_tokens": 18356 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the soft pairs mostly share only broad subsystems (tokenizers, loading, multimodal, device_map, training), not the same concrete bug. I would not merge any of the suggested duplicate edges.", + "confidence": 0.92, + "canonical_issue_reason": "No single issue in this set cleanly represents a duplicate cluster; the items cover unrelated bugs across training, loading, tokenization, multimodal processors, and device mapping.", + "canonical_pr_reason": null, + "best_issue_reason": "No best issue candidate is convincing because the cluster does not appear to describe one underlying problem.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both mention save/serialization, but one is tied lm_head serialization and the other is incorrect visual encoder keys; different bugs and code paths." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "One is a processor.tokenizer regression in v4.57.5; the other is a broader tokenization change. Not the same failure." + }, + { + "left": "issue:26413", + "right": "issue:28530", + "accept": false, + "reason": "Resume-from-checkpoint epoch iterator failure and early-stopping metric lookup are unrelated trainer issues." + }, + { + "left": "issue:44451", + "right": "issue:47020", + "accept": false, + "reason": "Both are model-loading/tokenizer-related, but they point to different models and different root causes." + }, + { + "left": "issue:43065", + "right": "issue:46402", + "accept": false, + "reason": "Both concern Sam3, but one is a dummy Conv2d in the pixel decoder while the other is missing text-encoder weight loading." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "Different models and different failure modes: torch.compile with SDPA vs accelerate loading of packed weights." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Both are device/runtime bugs, but one is SDPA compilation and the other is CPU offload device mismatch." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "Blip2 dtype propagation and Trainer max_steps on checkpoint resume are unrelated." + }, + { + "left": "issue:46821", + "right": "issue:46869", + "accept": false, + "reason": "TimesFM window_size AttributeError and YOLO multi-GPU device_map auto are different subsystems and bugs." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "Tokenizer crash on NER/padding is unrelated to OmDet-Turbo image-size mismatch." + }, + { + "left": "issue:43646", + "right": "issue:46665", + "accept": false, + "reason": "Both involve model initialization/loading, but one is a Transformers 5.0 breakage in custom init and the other is inconsistent from_pretrained behavior in Python classes." + }, + { + "left": "issue:46612", + "right": "issue:46858", + "accept": false, + "reason": "Both touch generation/cache, but one is skipped beam-search cache reorder and the other is a static-cache seq_length device-tensor arithmetic bug." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Different models and different parallelism problems: auto device_map CUDA assert vs tensor-parallel-only failure." + }, + { + "left": "issue:46897", + "right": "issue:47317", + "accept": false, + "reason": "Same general symptom class (training loss), but different models and no evidence of the same underlying bug." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Different Gemma models and different bugs: variable-name collision in forward vs PLE device mismatch." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "FineGrainedFP8Config failure and dequantized-model offload failure are distinct loading/configuration issues." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both affect serve/multimodal paths, but they fail in different places with different processor/model assumptions." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both are vision preprocessing issues, but one is a fixed-size processor/model mismatch and the other is non-default-resolution behavior with interpolate_pos_encoding." + }, + { + "left": "issue:28530", + "right": "issue:33666", + "accept": false, + "reason": "Early stopping metric resolution and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:46291", + "accept": false, + "reason": "FineGrainedFP8 loading and kernels import-version requirements are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Mistral3 tied-weight serialization and Qwen3.5 visual encoder key serialization are different bugs." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE and missing pad_token_idx in StableLmConfig are unrelated model/config issues." + }, + { + "left": "issue:45704", + "right": "issue:47404", + "accept": false, + "reason": "Apex RMSNorm memory leak and device_map loading OOM from conversion transients are different problems." + }, + { + "left": "issue:43716", + "right": "issue:46821", + "accept": false, + "reason": "Image preprocessor/model dtype mismatch and TimesFM window_size AttributeError do not share a code path." + }, + { + "left": "issue:28530", + "right": "issue:30064", + "accept": false, + "reason": "Trainer early stopping and image processor void-segmentation handling are unrelated." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Both are loading failures, but for different models with no shared root cause shown." + }, + { + "left": "issue:43441", + "right": "issue:47337", + "accept": false, + "reason": "FlashAttention failure for Ministral-3 and a flaky bbox-range test are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Different tokenizers, different tasks, and different AttributeError causes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 17, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:29159", + "issue:29925", + "issue:30064", + "issue:33357", + "issue:34567", + "issue:34634", + "issue:35707", + "issue:36032", + "issue:41720", + "issue:42490", + "issue:42915", + "issue:43257", + "issue:43316", + "issue:43322", + "issue:43329", + "issue:43452", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:44387", + "issue:44464", + "issue:44534", + "issue:44589", + "issue:44671", + "issue:44734", + "issue:44743", + "issue:44913", + "issue:45406", + "issue:45464", + "issue:45479", + "issue:45632", + "issue:45684", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:46665", + "issue:46821", + "issue:46858", + "issue:46894", + "issue:47020", + "issue:47254", + "issue:47317" + ], + "soft_pairs": [ + "issue:44734|issue:46858", + "issue:44464|issue:46858", + "issue:30064|issue:34634", + "issue:43452|issue:47020", + "issue:29159|issue:36032", + "issue:44671|issue:44743", + "issue:44387|issue:45701", + "issue:42915|issue:43322", + "issue:43873|issue:45704", + "issue:43531|issue:45759", + "issue:45479|issue:47317", + "issue:33357|issue:35707", + "issue:43540|issue:43782", + "issue:43550|issue:43653", + "issue:43316|issue:44913", + "issue:41720|issue:43746", + "issue:45632|issue:45684", + "issue:44534|issue:46665", + "issue:43606|issue:43720", + "issue:43257|issue:47254", + "issue:46894|issue:47317", + "issue:43531|issue:46821", + "issue:34634|issue:41720", + "issue:43531|issue:44589", + "issue:42490|issue:43329", + "issue:29159|issue:29925", + "issue:45406|issue:45464", + "issue:34567|issue:35707" + ], + "prepared_review_unit_hash": "5fc7ae71dcb6eb429a4f3182f0fdbc984455aec3fe159deaf02a022aeb9047f6", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36132, + "estimated_input_tokens": 9033, + "estimated_eval_tokens": 18322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft pairs mostly share broad subsystem words (tokenizer, loading, cache, Qwen, windowing) but not the same concrete bug or change. I would not merge any of the candidate pairs.", + "confidence": 0.94, + "canonical_issue_reason": "issue:29159 is the broadest and most representative issue in the set, centered on slow-vs-fast tokenizer behavioral inconsistency, which is the closest to the few tokenizer-related candidates.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:29159 is the best single issue to keep as a representative anchor because it has the most general problem statement and the widest semantic overlap with other tokenizer issues in the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44734", + "right": "issue:46858", + "accept": false, + "reason": "Both involve cache/compiled-generation failures, but one is KV-cache continuation indexing in serve and the other is static-cache sequence-length arithmetic on non-CUDA backends." + }, + { + "left": "issue:44464", + "right": "issue:46858", + "accept": false, + "reason": "Both concern generation under compile/cache, but the bugs are different: chunked generation output mismatch vs device-tensor seq-length breaking slice arithmetic." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Unrelated bugs in different areas: void segmentation map processing vs Bark voice_preset handling." + }, + { + "left": "issue:43452", + "right": "issue:47020", + "accept": false, + "reason": "Both are tokenizer-loading related, but one is gguf_file/from_pretrained path handling and the other is a constructor type mismatch in XLMRobertaTokenizer." + }, + { + "left": "issue:29159", + "right": "issue:36032", + "accept": false, + "reason": "Both are tokenizer issues, but one is slow/fast behavioral inconsistency while the other is a T5 class/method name conflict during loading." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "Different code paths and symptoms: incorrect CamemBERT masked-LM outputs vs recurrent state reset when caching with seq_len>1." + }, + { + "left": "issue:44387", + "right": "issue:45701", + "accept": false, + "reason": "Int4 quantization memory growth/OOM is unrelated to the tokenization changes described in the other issue." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "Both are model-load failures, but one is Qwen3Moe fine-grained FP8 config handling and the other is a Llava Next segmentation fault." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Different bugs: quantization/offloading behavior vs a T5 apex RMSNorm memory leak." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Both mention Qwen3, but one is sliding_window behavior in Qwen3-MoE and the other is config unwrapping for composite multimodal models." + }, + { + "left": "issue:45479", + "right": "issue:47317", + "accept": false, + "reason": "Both are training-loss problems, but there is no evidence they share the same loss computation bug or code path." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "A MacOS CLIP bus error is unrelated to progressive generation with inputs_embeds/past_key_values." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Both are Qwen multimodal issues, but one is a video-input processing ValueError and the other is a from_pretrained weight_only load error." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "Completely different subsystems: torch.compile/SDPA in Bamba vs BigBird tokenizer special-token registration." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Both involve config behavior, but one is an API discrepancy in Gemma3TextConfig and the other is rotary_pct not persisting on reload." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Different problems: device-mapping cudaErrorAssert on Qwen3 vs GraniteSpeech PEFT checkpoint loading." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "One is a trust_remote_code cache-path collision and the other is save_pretrained permission propagation; same area, different bug." + }, + { + "left": "issue:44534", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer initialization in v5 is unrelated to from_pretrained behavior inside Python classes." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "Both involve loading/offload/accelerate, but one is a device mismatch with CPU offload and the other is missing unpacking of packed BitNet weights." + }, + { + "left": "issue:43257", + "right": "issue:47254", + "accept": false, + "reason": "Both touch distributed training/loading, but the concrete failures are different: Qwen3 MOE weight conversion vs PEFT ZeRO-3 checkpointing." + }, + { + "left": "issue:46894", + "right": "issue:47317", + "accept": false, + "reason": "Both are training-loss bugs, but they refer to different models and no shared implementation defect is evident." + }, + { + "left": "issue:43531", + "right": "issue:46821", + "accept": false, + "reason": "Window-related names aside, sliding_window in Qwen3-MoE and TimesFM window_size attribute errors are different bugs." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "BarkProcessor voice_preset handling and Qwen3 auto device mapping failures are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "No shared bug: Qwen3-MoE sliding_window behavior vs Float8 storage lookup failure." + }, + { + "left": "issue:42490", + "right": "issue:43329", + "accept": false, + "reason": "Local-path trailing-slash import warning is unrelated to the multimodal video-token counting bug." + }, + { + "left": "issue:29159", + "right": "issue:29925", + "accept": false, + "reason": "Related theme, but not the same bug: generic slow/fast tokenizer inconsistency vs SigLIP specifically not forcing use_fast=True." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Different runtime failures in serve/chat APIs: missing _tokenizer on Gemma4Processor vs Qwen3.5 streaming inference crash." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token-count tracking and progressive generation with inputs_embeds/past_key_values are unrelated issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 18, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:42886", + "issue:43388", + "issue:43421", + "issue:43526", + "issue:43540", + "issue:43606", + "issue:43644", + "issue:43653", + "issue:43746", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:43927", + "issue:44062", + "issue:44077", + "issue:44451", + "issue:44464", + "issue:44492", + "issue:44509", + "issue:44514", + "issue:44545", + "issue:44610", + "issue:44625", + "issue:44843", + "issue:44898", + "issue:44918", + "issue:44936", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45059", + "issue:45072", + "issue:45357", + "issue:45479", + "issue:45704", + "issue:45759", + "issue:45859", + "issue:45923", + "issue:46050", + "issue:46402", + "issue:46665", + "issue:46858", + "issue:47050", + "issue:47333", + "issue:47405" + ], + "soft_pairs": [ + "issue:43881|issue:44991", + "issue:43606|issue:43653", + "issue:44492|issue:44509", + "issue:45759|issue:45859", + "issue:44451|issue:44991", + "issue:43540|issue:46858", + "issue:45704|issue:47405", + "issue:43526|issue:43883", + "issue:43388|issue:45059", + "issue:44987|issue:45042", + "issue:44077|issue:44625", + "issue:43644|issue:46665", + "issue:45923|issue:47333", + "issue:44898|issue:45072", + "issue:45357|issue:46402", + "issue:44062|issue:44451", + "issue:39692|issue:44610", + "issue:43746|issue:46050", + "issue:30064|issue:36010", + "issue:44918|issue:44936", + "issue:43746|issue:43873", + "issue:42886|issue:44843", + "issue:43421|issue:43927", + "issue:44625|issue:45479", + "issue:44514|issue:44545", + "issue:44464|issue:47050" + ], + "prepared_review_unit_hash": "2561c5b3e93d65e133264d2ab099ed734192930cbd7fc35674386eb4a55e2b7b", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35841, + "estimated_input_tokens": 8961, + "estimated_eval_tokens": 18178 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issue set, with two plausible duplicate pairs: an exact duplicate Qwen2_5_VLProcessor crash report, and a likely shared HF_HUB_OFFLINE tokenizer-cache regression. The remaining soft edges are only broad topical similarity, not the same concrete bug.", + "confidence": 0.79, + "canonical_issue_reason": "Issue 44514 is the cleanest issue anchor: it is the earlier report in an exact duplicate pair with the same title as 44545, and it contains enough reproduction detail to represent that bug well.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44514 is the strongest issue to keep as the representative bug report because it exactly matches 44545 and cleanly describes the batched apply_chat_template crash.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Both are model/tokenizer loading failures, but they concern different models and different failure modes; too broad to treat as the same bug." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "Completely different subsystems: CPU offload device mismatch vs tokenizer special-token decoding behavior." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "Unrelated: a cache-strategy typo versus documentation still mentioning removed pipeline tasks." + }, + { + "left": "issue:45759", + "right": "issue:45859", + "accept": false, + "reason": "Both involve Qwen 3.5/3.6, but one is a from_config/text_config unwrap issue and the other is a CPU-offload rotary embedding bug; different code paths." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both mention loading failures, but the models and likely root causes differ; not the same concrete tokenizer bug." + }, + { + "left": "issue:43540", + "right": "issue:46858", + "accept": false, + "reason": "Different features and failure surfaces: video input processing vs cache sequence length / torch.compile arithmetic." + }, + { + "left": "issue:45704", + "right": "issue:47405", + "accept": false, + "reason": "Unrelated problems: Apex RMSNorm memory leak in T5 versus wrong model class loading silently producing random weights." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "Different areas entirely: BeitImageProcessorFast label reduction vs Molmo tied-weights attribute error." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "Both involve vision/multimodal pipelines, but the symptoms and code paths are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "Different regressions: a specific model load failure versus PIL image processors incorrectly requiring torchvision." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "Not the same bug; one is a patchtsmixer post_init policy issue, the other is Qwen3.5 num_labels propagation." + }, + { + "left": "issue:43644", + "right": "issue:46665", + "accept": false, + "reason": "Both involve from_pretrained behavior, but one is about junk non-persistent buffers and the other is a Python-class usage inconsistency." + }, + { + "left": "issue:45923", + "right": "issue:47333", + "accept": false, + "reason": "Different features and failures: missing gradient_checkpointing support versus disk-offload save failure." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Different model families and failure modes: interpolation at non-default resolution vs bfloat16 dtype mismatch." + }, + { + "left": "issue:45357", + "right": "issue:46402", + "accept": false, + "reason": "Both are model save/load issues, but they affect different models and opposite directions (saving incorrect keys vs not loading weights)." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "Tokenizer-related, but the specific AddedToken constructor error is not the same as a generic latest-version load failure." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Both mention processor/model mismatch, but one is a SigLIP2 docs/example issue and the other is an OmDet-Turbo image-size mismatch." + }, + { + "left": "issue:43746", + "right": "issue:46050", + "accept": false, + "reason": "Different persistence bugs: PEFT adapter loading from local checkpoints versus quantized model save/load corruption." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Completely unrelated: segmentation-map processing versus an import error for GenerationMixin." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "Both touch trainers, but they describe different failure points and likely different root causes." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "Not the same bug; one is PEFT checkpoint loading, the other is a general offloading/quantization behavior issue." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": true, + "reason": "These describe the same HF_HUB_OFFLINE tokenizer-cache regression; 44843 provides the likely root cause in _patch_mistral_regex/model_info." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Different tokenizer/config bugs: runtime special-token post-processing versus DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "Different classification bugs with different scopes and symptoms; not the same underlying issue." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Exact same title and same batched apply_chat_template crash; these are duplicate reports of the same bug." + }, + { + "left": "issue:44464", + "right": "issue:47050", + "accept": false, + "reason": "Both mention generation/inference, but one is chunked generation with compiled forward and the other is an ONNX export dtype mismatch." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the accepted pairs are conservative and well-supported by the titles: 44514/44545 are an exact duplicate pair, and 42886/44843 plausibly describe the same offline tokenizer-cache bug. The remaining soft-edge rejections stay appropriately cautious about different models, subsystems, or failure modes." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 19, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:30064", + "issue:36331", + "issue:39692", + "issue:43295", + "issue:43381", + "issue:43421", + "issue:43493", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43643", + "issue:43653", + "issue:43673", + "issue:43720", + "issue:43782", + "issue:43819", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44368", + "issue:44488", + "issue:44661", + "issue:44671", + "issue:44821", + "issue:44857", + "issue:44995", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45356", + "issue:45357", + "issue:45381", + "issue:45834", + "issue:45907", + "issue:46032", + "issue:46294", + "issue:46463", + "issue:46655", + "issue:46693", + "issue:46858", + "issue:47020", + "issue:47050", + "issue:47337" + ], + "soft_pairs": [ + "issue:46655|issue:46693", + "issue:44995|issue:46858", + "issue:39692|issue:43493", + "issue:43381|issue:43844", + "issue:43421|issue:43653", + "issue:43673|issue:47337", + "issue:43782|issue:45381", + "issue:44857|issue:47050", + "issue:45081|issue:45356", + "issue:46032|issue:46463", + "issue:44368|issue:45357", + "issue:39692|issue:47050", + "issue:43421|issue:44186", + "issue:39692|issue:43873", + "issue:43994|issue:44671", + "issue:45356|issue:45907", + "issue:43720|issue:43819", + "issue:43596|issue:44220", + "issue:43540|issue:46294", + "issue:43550|issue:43819", + "issue:43550|issue:43927", + "issue:43295|issue:45834", + "issue:44821|issue:45084", + "issue:45072|issue:47050", + "issue:30064|issue:36331", + "issue:43653|issue:43720", + "issue:44062|issue:44488", + "issue:44488|issue:47020", + "issue:44220|issue:44661", + "issue:43531|issue:43643" + ], + "prepared_review_unit_hash": "145ff995c5f50678ad77afdc4794d4f15bec804e37951ae3485d98c5c66c00ce", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36992, + "estimated_input_tokens": 9248, + "estimated_eval_tokens": 18752 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft pair looks like a true duplicate; these are mostly unrelated issues across different models, tokenizer paths, generation, training, and documentation. The cluster does not have a clear canonical issue or PR.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue: the items span distinct subsystems and failure modes, and none of the soft pairs appears to describe the same underlying bug.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue: there is no cohesive duplicate cluster here, so any one issue would be an arbitrary representative rather than a true canonical.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46655", + "right": "issue:46693", + "accept": false, + "reason": "Video padding/calculation bug vs long-context FA2 performance regression; different code paths and symptoms." + }, + { + "left": "issue:44995", + "right": "issue:46858", + "accept": false, + "reason": "Stale indexer cache on second forward vs cache length tensor type in static-cache torch.compile; unrelated failures." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "SigLIP2 documentation/example errors vs HF/JAX implementation discrepancy; one is usage docs, the other is model behavior." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Eval-mode checkpointing restriction vs ZeRO-3 gradient inflation during training; not the same bug." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Runtime special-token/post-processor update bug vs BigBirdTokenizer mask-token registration bug; different tokenizer mechanics." + }, + { + "left": "issue:43673", + "right": "issue:47337", + "accept": false, + "reason": "Generation cache missing in chunked prefill vs flaky bbox-range test; unrelated areas." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Qwen3VL weight_only load error vs Qwen2.5-VL video position-id mismatch; different model families and problems." + }, + { + "left": "issue:44857", + "right": "issue:47050", + "accept": false, + "reason": "LwDetrImageLoss AMP crash vs CanineModel fp16 ONNX export dtype mismatch; separate model/export paths." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Mistral regex patch crash vs Kimi-K2.5 codec/regex regression; related tokenizer code, but distinct concrete failures." + }, + { + "left": "issue:46032", + "right": "issue:46463", + "accept": false, + "reason": "Mamba2 cache/use_cache incorrect results vs T5 decoder-input validation message; different subsystems and behaviors." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "Tie-word-embeddings warning in fine-tuning vs Qwen3.5 save_pretrained visual-key regression; not the same issue." + }, + { + "left": "issue:39692", + "right": "issue:47050", + "accept": false, + "reason": "SigLIP2 doc example bug vs CanineModel ONNX dtype mismatch; unrelated models and code paths." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Post-processor update at runtime vs LayoutLMv2 NER/padding crash; different tokenizer bugs." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 example/model mismatch vs quantization offloading behavior; no shared underlying defect." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "SigLIP2 AutoModel/pipeline issue vs CamemBERT masked LM regression; different models and failure modes." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regression warning vs list_repo_files revision forwarding in CI; unrelated functionality." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "BitNet packed-weights loading bug vs DAC missing-STE mismatch; different models and code paths." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Deepspeed Zero-3 BertModel init error vs audio fbank feature extraction issue; unrelated." + }, + { + "left": "issue:43540", + "right": "issue:46294", + "accept": false, + "reason": "Qwen3OmniMoe video processing ValueError vs Qwen3.5 video chat-template token mismatch; different model/input handling bugs." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Bamba torch.compile SDPA failure vs DAC latent/forward mismatch; unrelated models and defects." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Bamba compile/SDPA bug vs DiaConfig save/load token-id loss; not the same underlying problem." + }, + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Processor/tokenizer regression vs Kosmos2.5 long OCR index error; different code paths and model families." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor URL loading vs template compilation TypeError; unrelated failure modes." + }, + { + "left": "issue:45072", + "right": "issue:47050", + "accept": false, + "reason": "SwitchTransformers/TimmWrapper dtype mismatch vs CanineModel attention dtype mismatch; same broad symptom, but different models and fixes." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "ImageProcessor segmentation-map handling vs CustomTrainer compute_loss signature mismatch; unrelated components." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "Tokenizer special-token registration bug vs BitNet packed-weight loading bug; different modules and fixes." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken special-kwarg collision vs sleng-bert load failure; not the same bug." + }, + { + "left": "issue:44488", + "right": "issue:47020", + "accept": false, + "reason": "Model load failure for sleng-bert vs XLMRobertaTokenizer/Unigram constructor type mismatch; different tokenizer/model issues." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio fbank extraction issue vs add-new-model-like tokenizer-mapping failure; unrelated." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Qwen3-MoE sliding_window bug vs trust_remote_code AutoConfig missing fields; different code paths and symptoms." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 20, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:34634", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:43322", + "issue:43335", + "issue:43526", + "issue:43540", + "issue:43606", + "issue:43611", + "issue:43618", + "issue:43643", + "issue:43653", + "issue:43697", + "issue:43756", + "issue:43819", + "issue:43881", + "issue:44008", + "issue:44186", + "issue:44442", + "issue:44448", + "issue:44561", + "issue:44671", + "issue:44805", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44936", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45381", + "issue:45685", + "issue:45834", + "issue:45987", + "issue:46291", + "issue:46294", + "issue:46463", + "issue:46650", + "issue:46665", + "issue:46693", + "issue:46829", + "issue:46832", + "issue:46858", + "issue:46932", + "issue:47020", + "issue:47405" + ], + "soft_pairs": [ + "issue:41720|issue:44991", + "issue:45381|issue:46693", + "issue:43540|issue:44008", + "issue:43611|issue:46665", + "issue:43653|issue:43819", + "issue:44186|issue:44898", + "issue:46829|issue:46832", + "issue:34634|issue:36010", + "issue:46650|issue:47405", + "issue:44991|issue:47020", + "issue:43526|issue:43881", + "issue:44008|issue:46858", + "issue:44843|issue:45081", + "issue:44561|issue:45020", + "issue:39692|issue:45834", + "issue:45987|issue:46829", + "issue:43643|issue:44877", + "issue:43335|issue:43756", + "issue:44186|issue:44442", + "issue:44448|issue:44671", + "issue:45987|issue:46932", + "issue:44991|issue:46291", + "issue:46294|issue:46463", + "issue:44805|issue:44936", + "issue:43322|issue:45685", + "issue:43618|issue:43697", + "issue:43606|issue:45072" + ], + "prepared_review_unit_hash": "f6cad95fb364dc7bdce5a08a6c4e4e7280ba971a65b284ab457321b5ff12761f", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36226, + "estimated_input_tokens": 9057, + "estimated_eval_tokens": 18370 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These issues span many unrelated models, tokenizers, configs, and backend/device paths. None of the proposed soft pairs look like true duplicates or should be merged into one PR/issue.", + "confidence": 0.99, + "canonical_issue_reason": "No single canonical issue fits this cluster: the items are heterogeneous and do not share one underlying bug or code path.", + "canonical_pr_reason": null, + "best_issue_reason": "No clear global best issue. The closest broad umbrella is 45020 for remote_code regressions, but it does not represent the many unrelated failures here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Different failures: Qwen3 auto device mapping CUDA assert vs tokenizer loading for EMBEDDIA/est-roberta." + }, + { + "left": "issue:45381", + "right": "issue:46693", + "accept": false, + "reason": "Different code paths: Qwen2.5-VL video position IDs vs FA2 long-context performance regression." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Different models and bugs: Qwen3OmniMoe video ValueError vs Gemma3n forward variable collision." + }, + { + "left": "issue:43611", + "right": "issue:46665", + "accept": false, + "reason": "Both involve loading APIs, but the root causes differ: base_model_prefix breakage vs inconsistent from_pretrained inside classes." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Unrelated subsystems: BigBirdTokenizer special-token decoding vs DAC.from_latents mismatch." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "Different models and symptoms: LayoutLMv2Tokenizer NER/padding crash vs Perceiver image interpolation failure." + }, + { + "left": "issue:46829", + "right": "issue:46832", + "accept": false, + "reason": "Different RT-DETR/VideoPrism bugs: output field mismatch vs feature-level type error." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "Processor preset handling and a GenerationMixin import regression are unrelated bugs." + }, + { + "left": "issue:46650", + "right": "issue:47405", + "accept": false, + "reason": "Both touch model loading/saving, but one is reverse-rename corruption and the other is silent wrong-auto-class conversion skipping." + }, + { + "left": "issue:44991", + "right": "issue:47020", + "accept": false, + "reason": "Different tokenizer implementations and failure modes; no shared underlying bug." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug vs glm-4v-9b loading failure are unrelated." + }, + { + "left": "issue:44008", + "right": "issue:46858", + "accept": false, + "reason": "Gemma3n attribute collision and static-cache seq-length tensor arithmetic are separate code paths." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Same helper name, different bug: offline model_info call vs backend_tokenizer AttributeError." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Both mention trust_remote_code, but they are different regressions with different causes." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "Docs-example issues for SigLIP2 vs a long-OCR runtime error in Kosmos2.5 are not the same bug." + }, + { + "left": "issue:45987", + "right": "issue:46829", + "accept": false, + "reason": "StaticCache shape contract issue vs VideoPrism output misassignment are unrelated." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "Both are config/loading problems, but missing fields from trust_remote_code and strict granite_speech config are different issues." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "Different architectures and failure modes: sparse encoder layer creation vs RoPE layer dropping." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "Tokenizer crashes, but the affected tokenizers and root causes differ." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Model-specific v4/v5 behavior regressions, but not the same model or code path." + }, + { + "left": "issue:45987", + "right": "issue:46932", + "accept": false, + "reason": "Static cache length contract vs Gemma4 quantized embedder dtype casting are unrelated." + }, + { + "left": "issue:44991", + "right": "issue:46291", + "accept": false, + "reason": "Tokenizer loading failure and kernels import-time dependency failure are separate problems." + }, + { + "left": "issue:46294", + "right": "issue:46463", + "accept": false, + "reason": "Qwen video chat-template mismatch and T5 decoder validation messaging are unrelated." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "A tensor mask shape mismatch and trainer.evaluate() failing after train are not the same underlying issue." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Different backends/models: Llava Next segmentation fault vs MPS histogram support error." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "CLIPOutput attentions regression and RTDetrV2 output differences are unrelated model-specific bugs." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "CPU offload device mismatch and bfloat16 dtype mismatch are different failure modes in different models." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 21, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28218", + "issue:29159", + "issue:34634", + "issue:36331", + "issue:42175", + "issue:42915", + "issue:43388", + "issue:43421", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43782", + "issue:43818", + "issue:43825", + "issue:43931", + "issue:43994", + "issue:44008", + "issue:44291", + "issue:44479", + "issue:44610", + "issue:44811", + "issue:44977", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45183", + "issue:45381", + "issue:45479", + "issue:45563", + "issue:45584", + "issue:45834", + "issue:46032", + "issue:46512", + "issue:46665", + "issue:46752", + "issue:46846", + "issue:47254", + "issue:47276", + "issue:47333", + "issue:47337" + ], + "soft_pairs": [ + "issue:43716|issue:44008", + "issue:45584|issue:45834", + "issue:43550|issue:45072", + "issue:44479|issue:45834", + "issue:43421|issue:44811", + "issue:45183|issue:45479", + "issue:46752|issue:47337", + "issue:43782|issue:44977", + "issue:43644|issue:43749", + "issue:43577|issue:43818", + "issue:43582|issue:46512", + "issue:45563|issue:47276", + "issue:44610|issue:45072", + "issue:34634|issue:36331", + "issue:43720|issue:45072", + "issue:28218|issue:29159", + "issue:43540|issue:45381", + "issue:43645|issue:46665", + "issue:43931|issue:46846", + "issue:43749|issue:44291", + "issue:47276|issue:47337", + "issue:43742|issue:45020", + "issue:43825|issue:43994", + "issue:42175|issue:45081", + "issue:47254|issue:47333", + "issue:43388|issue:46032", + "issue:42915|issue:47333" + ], + "prepared_review_unit_hash": "0db526fec7064d439f877e3a715ff000fab3acd41cd76cb1868f672ec855dd75", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35531, + "estimated_input_tokens": 8883, + "estimated_eval_tokens": 18022 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab bag of largely unrelated transformer bugs (tokenizer, multimodal model loading, dtype/loading regressions, trainer/runtime issues). The soft similarities are mostly thematic, not the same underlying defect, so none of the proposed duplicate pairs should be merged.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Both involve multimodal models, but one is a preprocessor/model dtype mismatch and the other is a Gemma3n forward variable-collision AttributeError; different code paths and fixes." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty-transcription failure after special-token alignment is unrelated to Kosmos2.5 long-OCR index handling." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "One is a Bamba torch.compile/SDPA bug, the other is a dtype mismatch in different models; not the same concrete failure." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "Qwen VL video-input regression and Kosmos2.5 long OCR index error are different model/input bugs." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Both are tokenizer-adjacent, but runtime post-processor updates and Whisper batch_decode skip_special_tokens are separate behaviors." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Input validation XOR messaging is unrelated to the single-label-classification zero-loss bug." + }, + { + "left": "issue:46752", + "right": "issue:47337", + "accept": false, + "reason": "Empty chat-template input crash and bbox-range flaky test are unrelated." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Qwen3VL weight_only loading error and Qwen3.5 flash-attention generation failure are different issues." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Both concern loading, but junk non-persistent buffers and broken FSDP CPU RAM efficient loading are distinct defects." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "BLIP2 dtype not propagating is unrelated to Video-LLaVA missing temporal attention/shared weights." + }, + { + "left": "issue:43582", + "right": "issue:46512", + "accept": false, + "reason": "Apple Silicon caching_allocator warmup TypeError and stale model.config.dtype after from_config are different dtype-related bugs." + }, + { + "left": "issue:45563", + "right": "issue:47276", + "accept": false, + "reason": "Both are warning-message problems, but about different APIs and different message text." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "OmDet-Turbo processor size mismatch is unrelated to SwitchTransformers/TimmWrapperModel bfloat16 dtype mismatches." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "BarkProcessor voice_preset and CustomTrainer.compute_loss signature mismatch are unrelated." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "BitNet packed-weight unpacking during loading and a separate dtype mismatch in inference are not the same bug." + }, + { + "left": "issue:28218", + "right": "issue:29159", + "accept": false, + "reason": "Tokenizer extra-space after added token is a specific bug; the other is a broader slow/fast tokenizer inconsistency report." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Qwen3OmniMoe video-input ValueError and generic input_ids/inputs_embeds validation messaging are unrelated." + }, + { + "left": "issue:43645", + "right": "issue:46665", + "accept": false, + "reason": "Jupyter custom-model initialization breakage and from_pretrained behavior inside Python classes are different loading scenarios with different causes." + }, + { + "left": "issue:43931", + "right": "issue:46846", + "accept": false, + "reason": "Different Qwen variants and different failure modes: weight shape mismatch versus missing tp-plan entries causing OOM/shape errors." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and init_empty_weights unexpected argument are separate loading regressions." + }, + { + "left": "issue:47276", + "right": "issue:47337", + "accept": false, + "reason": "Misleading synced_gpus warning and bbox coordinate range flaky test are unrelated." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "A key error loading one model is not the same underlying problem as broad remote_code breakage across recent versions." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "Wrong translation-support warning text and SigLIP2 nonsensical outputs are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Package dependency omission and a tokenizer backend_tokenizer AttributeError are unrelated." + }, + { + "left": "issue:47254", + "right": "issue:47333", + "accept": false, + "reason": "PEFT+ZeRO-3 checkpointing error and disk-offloaded Qwen3 save failure are distinct training/offloading bugs." + }, + { + "left": "issue:42915", + "right": "issue:47333", + "accept": false, + "reason": "FineGrainedFP8Config failure for Qwen3Moe and disk-offload save failure for Qwen3-30B-A3B are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 22, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:36010", + "issue:36331", + "issue:42175", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43452", + "issue:43531", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43638", + "issue:43645", + "issue:43653", + "issue:43782", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43927", + "issue:44060", + "issue:44368", + "issue:44589", + "issue:44671", + "issue:44821", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44991", + "issue:45003", + "issue:45042", + "issue:45081", + "issue:45325", + "issue:45372", + "issue:45406", + "issue:45701", + "issue:46175", + "issue:46285", + "issue:46291", + "issue:46326", + "issue:46697", + "issue:46728", + "issue:46851", + "issue:46932", + "issue:47217", + "issue:47254", + "issue:47332", + "issue:47337" + ], + "soft_pairs": [ + "issue:46697|issue:46728", + "issue:44671|issue:45701", + "issue:44589|issue:45042", + "issue:43577|issue:43582", + "issue:47254|issue:47332", + "issue:43782|issue:45325", + "issue:43262|issue:47217", + "issue:36010|issue:43645", + "issue:46851|issue:47337", + "issue:43262|issue:43819", + "issue:43653|issue:44912", + "issue:43582|issue:44857", + "issue:36010|issue:36331", + "issue:43531|issue:45042", + "issue:45003|issue:46285", + "issue:46175|issue:46932", + "issue:43550|issue:44912", + "issue:42175|issue:46291", + "issue:45372|issue:46326", + "issue:43638|issue:44936", + "issue:43819|issue:43927", + "issue:43825|issue:43827", + "issue:43257|issue:43329", + "issue:43452|issue:44991", + "issue:42915|issue:44821", + "issue:45081|issue:45406", + "issue:44060|issue:44368" + ], + "prepared_review_unit_hash": "71758b04dfe1d4a5b6a18af370e77563fcb180863923ae9fd0d720af70870a93", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36290, + "estimated_input_tokens": 9073, + "estimated_eval_tokens": 18402 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a bag of unrelated closed issues that only share broad lexical themes (v5 regressions, Qwen/Gemma/Gemma4, tokenizer/model loading, multimodal bugs). None of the soft-edge pairs look like the same underlying defect, so they should all stay separate.", + "confidence": 0.96, + "canonical_issue_reason": "If a single issue must anchor the cluster, issue:45701 is the broadest/highest-signal user-facing regression report here (multiple references, general tokenization breakage). But it is not a true duplicate of the others.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45701 is the most representative general regression report in this set; the rest are narrower, model-specific, or subsystem-specific bugs with no strong duplicate chain.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46697", + "right": "issue:46728", + "accept": false, + "reason": "Both are image-processor bugs, but one is over-padding in Swin2SR and the other is a wrong patch-count calculation in Aria; different code paths and symptoms." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "One reports incorrect CamemBERT masked-LM predictions, the other a version-dependent tokenizer change; related theme, not the same bug." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Float8 storage lookup failure and PIL backend requiring torchvision are unrelated runtime failures." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "BLIP2 dtype staying float32 and Apple Silicon warmup TypeError are different models/platforms and different failure modes." + }, + { + "left": "issue:47254", + "right": "issue:47332", + "accept": false, + "reason": "PEFT+ZeRO-3 checkpointing error and offloading-save failure for Qwen3-VL are separate training vs serialization problems." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Qwen3VL weight_only loading error and Qwen2.5-VL rope-index scaling bug are distinct model/version issues." + }, + { + "left": "issue:43262", + "right": "issue:47217", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and VL image-placeholder handling are different modalities and different code paths." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import failure and custom-model initialization in notebooks are unrelated import/environment regressions." + }, + { + "left": "issue:46851", + "right": "issue:47337", + "accept": false, + "reason": "Variable-length position_ids in Qwen3.5 and bbox range flakiness are unrelated bugs." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio processor chat-template sampling rate and DAC latent reconstruction mismatch are unrelated issues." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and MXFP4 quantization fallback are unrelated tokenizer vs quantization bugs." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError and float16 AMP crash in LwDetrImageLoss are different platform/model failures." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "Importing GenerationMixin and CustomTrainer.compute_loss signature mismatch are unrelated API breakages." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior and PIL backend torchvision dependency are unrelated." + }, + { + "left": "issue:45003", + "right": "issue:46285", + "accept": false, + "reason": "Unsafe sys.modules access in modeling_utils and CLIP text-model structure inconsistency are different internal API issues." + }, + { + "left": "issue:46175", + "right": "issue:46932", + "accept": false, + "reason": "The first is a memory blow-up in Gemma4 vision patch embeddings; the second is a dtype-casting bug for quantized embedders." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "Bamba torch.compile SDPA failure and MXFP4 quantization fallback are unrelated problems." + }, + { + "left": "issue:42175", + "right": "issue:46291", + "accept": false, + "reason": "Missing TensorFlow in torch-only install and kernels version/revision import failure are unrelated packaging issues." + }, + { + "left": "issue:45372", + "right": "issue:46326", + "accept": false, + "reason": "Both involve mistral_common-adjacent processor/tokenizer code, but one is an ImportError for ReasoningEffort and the other is continue_final_message behavior; not the same defect." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "ZeRO-3 index error with a non-pretrained Bert model and trainer.evaluate() failing after train() are unrelated training-loop issues." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "DAC latent reconstruction mismatch and DiaConfig losing custom token IDs are different components and different failure modes." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "One is an error-message regression about translation support in pipeline(), the other is stale docs still referencing pipeline(); related theme, not one bug." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed and a multimodal token-counting undefined-name bug are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "gguf_file breakage for AutoTokenizer/AutoModel loading and a specific tokenizer-loading failure for EMBEDDIA/est-roberta are separate pretrained-loading problems." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and AutoImageProcessor-from-URL failure are unrelated." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Mistral regex patching on a tokenizer without backend_tokenizer and Gemma4Processor missing _tokenizer are distinct tokenizer/processor bugs." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention tied-weights warnings in Qwen-family models, but they describe different models and different causes; not the same concrete bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 23, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28370", + "issue:29159", + "issue:34567", + "issue:38175", + "issue:39692", + "issue:40990", + "issue:42490", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43425", + "issue:43450", + "issue:43525", + "issue:43526", + "issue:43643", + "issue:43696", + "issue:43697", + "issue:43742", + "issue:43746", + "issue:43854", + "issue:43906", + "issue:43931", + "issue:44291", + "issue:44387", + "issue:44479", + "issue:44554", + "issue:44589", + "issue:44610", + "issue:44617", + "issue:44811", + "issue:44821", + "issue:44912", + "issue:44977", + "issue:44987", + "issue:45042", + "issue:45081", + "issue:45084", + "issue:45538", + "issue:45600", + "issue:45759", + "issue:45859", + "issue:46209", + "issue:46291", + "issue:47020", + "issue:47050", + "issue:47145", + "issue:47333" + ], + "soft_pairs": [ + "issue:43526|issue:43697", + "issue:43257|issue:43696", + "issue:43854|issue:44977", + "issue:45859|issue:47050", + "issue:28370|issue:29159", + "issue:43746|issue:44811", + "issue:42915|issue:43742", + "issue:45084|issue:45538", + "issue:38175|issue:44610", + "issue:43906|issue:46209", + "issue:43329|issue:44479", + "issue:43262|issue:44811", + "issue:44291|issue:45042", + "issue:42490|issue:43257", + "issue:43643|issue:45759", + "issue:43329|issue:45600", + "issue:43329|issue:43450", + "issue:39692|issue:44912", + "issue:34567|issue:40990", + "issue:43526|issue:44554", + "issue:44387|issue:44617", + "issue:44589|issue:47333", + "issue:29159|issue:34567", + "issue:43257|issue:43931", + "issue:45081|issue:47020", + "issue:43425|issue:46291", + "issue:44987|issue:45538", + "issue:44291|issue:44821", + "issue:43525|issue:46291", + "issue:44912|issue:47145" + ], + "prepared_review_unit_hash": "76d3e91e479b3d6b3078f0bc6cc118c9e0b804292b87b75200bf32e723e5f58d", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36781, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab-bag of unrelated Transformers issues that only share broad themes like loading, tokenizers, multimodal/video, quantization, or OOMs. None of the soft pairs look like the same underlying bug or a safe PR merge candidate.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "Different models and code paths: BeitImageProcessorFast label reduction vs RTDetrV2 output drift." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "One is Qwen3 MOE weight conversion with accelerate/deepspeed; the other is a GPT-oss CUDA OOM. Same broad area, not the same defect." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "Different models and symptoms: GLM-4.7-Flash loading in tests vs Qwen3.5 flash-attention generation problems." + }, + { + "left": "issue:45859", + "right": "issue:47050", + "accept": false, + "reason": "Unrelated failures: CPU-offload compatibility in Qwen3.5Moe rotary embedding vs CanineModel ONNX dtype mismatch." + }, + { + "left": "issue:28370", + "right": "issue:29159", + "accept": false, + "reason": "Both tokenizer-related, but one is missing vocab_file for custom SentencePiece and the other is slow/fast tokenizer inconsistency." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "Different functionality: GraniteSpeech local PEFT checkpoint loading vs Whisper batch_decode skip_special_tokens handling." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "Different model/config bugs: Qwen3Moe FineGrainedFP8Config failure vs MobileLLM-125M key error." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "No shared root cause: template compilation error vs CLIPTokenizer model_max_length being too large." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Different models and failure modes: SigLIP2 zero probabilities vs OmDet-Turbo processor/model image-size mismatch." + }, + { + "left": "issue:43906", + "right": "issue:46209", + "accept": false, + "reason": "Too generic to unify: an isolated reproduction of an earlier issue vs an FP8 inference failure on B200 hardware." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Both touch video/multimodal handling, but one is undefined symbols in _get_num_multimodal_tokens and the other is a broader v5.3.0 video regression across several models." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Different audio bugs: apply_chat_template default sampling rate vs Whisper decode ignoring skip_special_tokens." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "Different subsystems: init_empty_weights argument handling vs PIL backend image processors wrongly requiring torchvision." + }, + { + "left": "issue:42490", + "right": "issue:43257", + "accept": false, + "reason": "Local-path slash import warning and Qwen3 MOE weight-conversion failure are unrelated." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Both are config-loading issues, but the concrete bugs differ: missing fields under trust_remote_code vs not unwrapping text_config in composite multimodal configs." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Video token counting bug vs stale auto_mappings references to removed Sam3LiteText configs; different files and causes." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Both mention video, but one is undefined helper usage in token counting and the other is batched video processor shape output." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "Different problem domains: SigLIP2 documentation errors vs GPT-oss MXFP4 quantization fallback to bf16." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "Trainer token accounting bug vs extreme perplexity on gpt-oss; no shared underlying code path." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "Image processor label reduction bug and MPS attention correctness issue are unrelated." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Both can surface as OOM, but one is int4 quantization memory reservation and the other is a Sam3Video runtime OOM." + }, + { + "left": "issue:44589", + "right": "issue:47333", + "accept": false, + "reason": "Storage lookup failure for FP8 tensors vs saving a disk-offloaded Qwen3 model; not the same bug." + }, + { + "left": "issue:29159", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer slow/fast inconsistency is unrelated to TrainerState num_input_tokens_seen not updating." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Same model family name, but different failures: weight conversion with accelerate/deepspeed vs Qwen3-VL weight-shape mismatch on load." + }, + { + "left": "issue:45081", + "right": "issue:47020", + "accept": false, + "reason": "Both are tokenizer construction issues, but they involve different tokenizers and different failure mechanisms." + }, + { + "left": "issue:43425", + "right": "issue:46291", + "accept": false, + "reason": "Unrelated dependency problems: Torch 2.10 incompatibility vs kernels v0.15 import requirement." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Different components and bugs: loading physical-intelligence/fast on transformers>=5.1.0 vs CLIPTokenizer max-length misconfiguration." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "init_empty_weights TypeError and AutoImageProcessor URL loading are separate issues." + }, + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Llama4Config missing pad_token_id is unrelated to kernels-v0.15 pipeline import failure." + }, + { + "left": "issue:44912", + "right": "issue:47145", + "accept": false, + "reason": "Both are quantization-related, but one is MXFP4 falling back to bf16 and the other is silent corruption from missing weight_scale on frozen FP8 checkpoints." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 24, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:29159", + "issue:29870", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:42175", + "issue:42915", + "issue:43404", + "issue:43454", + "issue:43531", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43650", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43824", + "issue:43906", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44568", + "issue:44589", + "issue:44756", + "issue:44821", + "issue:44936", + "issue:44991", + "issue:45127", + "issue:45137", + "issue:45356", + "issue:45357", + "issue:45405", + "issue:45459", + "issue:45538", + "issue:45803", + "issue:46421", + "issue:46424", + "issue:46693", + "issue:46701", + "issue:46821", + "issue:46869", + "issue:47254", + "issue:47332" + ], + "soft_pairs": [ + "issue:45459|issue:45803", + "issue:43742|issue:44291", + "issue:43540|issue:46701", + "issue:28282|issue:34689", + "issue:28282|issue:29870", + "issue:43454|issue:43577", + "issue:33357|issue:34567", + "issue:43749|issue:44756", + "issue:44589|issue:47332", + "issue:43650|issue:43906", + "issue:42175|issue:43824", + "issue:43531|issue:44821", + "issue:44568|issue:45356", + "issue:44479|issue:46693", + "issue:46421|issue:46424", + "issue:42915|issue:45538", + "issue:24643|issue:29159", + "issue:43582|issue:47254", + "issue:44936|issue:46821", + "issue:45137|issue:45405", + "issue:43404|issue:45127", + "issue:28282|issue:29127", + "issue:43697|issue:45357", + "issue:44062|issue:44991", + "issue:45538|issue:46869" + ], + "prepared_review_unit_hash": "88fec16a666bba52edee3f9688b8056827c27631145ee4ca9bd371603f1314ff", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 25, + "serialized_chars": 34614, + "estimated_input_tokens": 8654, + "estimated_eval_tokens": 17564 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is dominated by superficial text similarity rather than true duplicates. The paired issues mostly span unrelated subsystems or root causes: tokenizer errors, multimodal model loading, cache behavior, DeepSpeed/FSDP, and model-specific regressions. I would not merge any of the soft pairs.", + "confidence": 0.89, + "canonical_issue_reason": "No clear duplicate hub emerges from the cluster; the issues are too heterogeneous. If one must be picked as a representative, issue 28282 is the clearest standalone user-facing failure report, but it does not unify the rest of the set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 28282 is the most concise and broadly understandable report, with an explicit error message and a clear reproduction context. It is representative as a high-signal issue, but not as a duplicate anchor for the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both involve error handling, but they hit different code paths and different failure modes; not the same underlying bug." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Both are model-loading failures, but one is a missing key and the other is an init_empty_weights argument mismatch; distinct causes." + }, + { + "left": "issue:43540", + "right": "issue:46701", + "accept": false, + "reason": "Different multimodal pipelines and different failure surfaces; only broad subsystem overlap." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "One is an environment/dependency import error, the other a model-loading regression for a specific vision model." + }, + { + "left": "issue:28282", + "right": "issue:29870", + "accept": false, + "reason": "Completely different bug classes: missing PyTorch vs decoder attention mask assignment in generation." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "Both touch model internals, but one is weight tying for a specific architecture and the other is dtype propagation on BLIP2 loading." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "Unrelated areas: a MacOS CLIP bus error versus a TrainerState bookkeeping bug." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "Both relate to memory/loading, but one is FSDP CPU RAM-efficient loading and the other is mmap/OOM on a specific machine family." + }, + { + "left": "issue:44589", + "right": "issue:47332", + "accept": false, + "reason": "Different storage/offload failures; no shared code-path is evident." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "No meaningful technical overlap beyond both being issue reports." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "Package/backend installation problem versus missing exported class import; not the same defect." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Different model families and unrelated failures; one is sliding_window behavior, the other image processor loading from URL." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." + }, + { + "left": "issue:44479", + "right": "issue:46693", + "accept": false, + "reason": "One is a video-input regression across Qwen VL variants; the other is a performance regression with FA2 and long-context generation." + }, + { + "left": "issue:46421", + "right": "issue:46424", + "accept": false, + "reason": "Both mention cache, but one is lazy-init causing recompilation and the other is max_cache_length being ignored; distinct bugs." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "No real overlap: FP8 configuration failure versus CLIP tokenizer max_length behavior." + }, + { + "left": "issue:24643", + "right": "issue:29159", + "accept": false, + "reason": "Different domains entirely: DeepSpeed training weight-shape error versus tokenizer slow/fast inconsistency." + }, + { + "left": "issue:43582", + "right": "issue:47254", + "accept": false, + "reason": "Both touch distributed or Apple/PEFT contexts, but the actual failures and code paths are unrelated." + }, + { + "left": "issue:44936", + "right": "issue:46821", + "accept": false, + "reason": "Trainer state/eval sequencing bug versus a TimesFM window_size attribute error; not duplicates." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO-3 deque failure and an unreleased PEFT version bump are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Both involve tied embeddings/lm_head, but one is a missing weight-tie implementation and the other is a LoRA merge with extended vocabulary; different root causes." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Missing PyTorch import versus LayoutLMv3 box-validation messaging; no duplicate evidence." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "One is an inference-output regression, the other a save_pretrained key serialization bug; different behaviors and fixes." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Tokenizer-related failures, but one is an AddedToken argument conflict and the other is a model-specific tokenizer loading regression." + }, + { + "left": "issue:45538", + "right": "issue:46869", + "accept": false, + "reason": "CLIP tokenizer max_length bug and YOLO multi-GPU device-map failure are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 25, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:27957", + "issue:28282", + "issue:28370", + "issue:34567", + "issue:36331", + "issue:38175", + "issue:39379", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43452", + "issue:43493", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43716", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43994", + "issue:44062", + "issue:44479", + "issue:44488", + "issue:44554", + "issue:44560", + "issue:44568", + "issue:44589", + "issue:44792", + "issue:44849", + "issue:44857", + "issue:44898", + "issue:44964", + "issue:44987", + "issue:45042", + "issue:45081", + "issue:45325", + "issue:45381", + "issue:45488", + "issue:45701", + "issue:45823", + "issue:45907", + "issue:46463", + "issue:46512", + "issue:46701", + "issue:46832", + "issue:47332", + "issue:47333" + ], + "soft_pairs": [ + "issue:44857|issue:46832", + "issue:41720|issue:43540", + "issue:43540|issue:44479", + "issue:39379|issue:44554", + "issue:44568|issue:45488", + "issue:43540|issue:44560", + "issue:42915|issue:47332", + "issue:45042|issue:45081", + "issue:42915|issue:43749", + "issue:27957|issue:28282", + "issue:43452|issue:43582", + "issue:34567|issue:36331", + "issue:44479|issue:45381", + "issue:44062|issue:45081", + "issue:43742|issue:44589", + "issue:42175|issue:44488", + "issue:38175|issue:44898", + "issue:43493|issue:43994", + "issue:43577|issue:46512", + "issue:28282|issue:28370", + "issue:44792|issue:44964", + "issue:43742|issue:47333", + "issue:44849|issue:45701", + "issue:44987|issue:45907", + "issue:43716|issue:46701", + "issue:45823|issue:46463", + "issue:43756|issue:45325" + ], + "prepared_review_unit_hash": "19fd2a16fe5486438065d56d76c03eb5a3b5d4cf7241364338deb43a7bb24c90", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35219, + "estimated_input_tokens": 8805, + "estimated_eval_tokens": 17866 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a set of unrelated model-specific bugs, with a clear Qwen video-input regression subgroup centered on issue:44479. A SigLIP2-related subgroup also appears, but there is not enough evidence to promote any PR canonical here.", + "confidence": 0.81, + "canonical_issue_reason": "issue:44479 is the broadest and most central report in the clearest duplicate family, covering the shared Qwen video-input regression across several related model variants.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 best represents the cluster because it anchors the most coherent duplicate subgroup while the rest of the issues are largely fragmented and unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44857", + "right": "issue:46832", + "accept": false, + "reason": "Different detector models and different failure modes; both are vision bugs but not the same underlying issue." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Both mention Qwen3, but one is an auto device-mapping CUDA assert and the other is a video-input processing error." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Related video-input failures in nearby Qwen model lines, but the packet does not show enough evidence that they are the same regression family." + }, + { + "left": "issue:39379", + "right": "issue:44554", + "accept": false, + "reason": "Different code paths: mllama mask handling versus MPS attention correctness." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Separate tokenizer bugs affecting different models and different tokenizer layers." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Both are video-related, but they hit different Qwen model families and error types; not clearly the same bug." + }, + { + "left": "issue:42915", + "right": "issue:47332", + "accept": false, + "reason": "Unrelated offloading/save issue versus a FineGrainedFP8Config failure." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Completely different subsystems: image processor dependency handling versus Mistral tokenizer regex patching." + }, + { + "left": "issue:42915", + "right": "issue:43749", + "accept": false, + "reason": "Different problems in Qwen/FP8 versus FSDP CPU RAM efficient loading." + }, + { + "left": "issue:27957", + "right": "issue:28282", + "accept": false, + "reason": "Unrelated issues: Flash Attention support versus missing PyTorch import error." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "Different failures in tokenizer/model loading versus Apple Silicon allocator warmup." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "Trainer token-count tracking and custom compute_loss signature are distinct trainer API issues." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Both describe the same Qwen2.5-VL/Qwen video-position regression in v5.3.0." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer AddedToken argument conflict is unrelated to the Mistral regex/backend_tokenizer crash." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "Different loading errors with no shared concrete code path." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Package/backend installation issue versus a model loading failure." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Different models and symptoms: zero probabilities in SigLIP2 versus Perceiver non-default resolution failure." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 family and same underlying implementation discrepancy producing incorrect outputs." + }, + { + "left": "issue:43577", + "right": "issue:46512", + "accept": false, + "reason": "Both involve dtype, but one is BLIP2/qformer dtype loading and the other is stale config.dtype from from_config; not the same bug." + }, + { + "left": "issue:28282", + "right": "issue:28370", + "accept": false, + "reason": "Missing PyTorch import error and missing vocab_file for custom SentencePiece models are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Different multimodal model failures with no clear shared concrete code-path bug." + }, + { + "left": "issue:43742", + "right": "issue:47333", + "accept": false, + "reason": "Different model loading versus disk-offloaded save failure." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "Qwen3.5 hidden-states bug and a tokenization-version change are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "Different repository/package issues; only loosely both mention versioning." + }, + { + "left": "issue:43716", + "right": "issue:46701", + "accept": false, + "reason": "Mistral-3 image dtype mismatch and DiffusionGemma pipeline failure are unrelated." + }, + { + "left": "issue:45823", + "right": "issue:46463", + "accept": false, + "reason": "Different subsystems: Gemma4 device mismatch versus T5 error messaging." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen/Smollm RoPE-positioning bugs with no clear shared underlying change." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary is mostly grounded, and issue:44479 is a reasonable canonical representative. However, at least one soft-edge verdict is a bit too aggressive: issue:43540 vs issue:44479 is not clearly the same bug from the packet alone. The titles show related video-input failures in different Qwen model lines, but not enough evidence to call them the same regression family with confidence. I\u2019d keep that pair conservative/uncertain rather than accept. The rest of the verdicts look more defensible." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 26, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:24643", + "issue:29870", + "issue:41762", + "issue:42796", + "issue:43257", + "issue:43452", + "issue:43540", + "issue:43577", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43756", + "issue:43906", + "issue:44361", + "issue:44466", + "issue:44488", + "issue:44568", + "issue:44625", + "issue:44737", + "issue:44756", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44936", + "issue:45005", + "issue:45072", + "issue:45405", + "issue:45588", + "issue:45600", + "issue:45663", + "issue:45672", + "issue:45674", + "issue:46121", + "issue:46399", + "issue:46463", + "issue:46491", + "issue:46693", + "issue:46821", + "issue:46829", + "issue:46846", + "issue:46899", + "issue:47276", + "issue:47332", + "issue:47404", + "issue:47405" + ], + "soft_pairs": [ + "issue:45672|issue:45674", + "issue:43756|issue:46399", + "issue:46463|issue:47276", + "issue:44737|issue:46463", + "issue:43257|issue:43756", + "issue:45405|issue:45600", + "issue:44625|issue:44877", + "issue:43257|issue:47405", + "issue:43742|issue:47332", + "issue:43577|issue:45072", + "issue:43540|issue:46693", + "issue:43906|issue:46693", + "issue:44466|issue:45005", + "issue:43452|issue:44488", + "issue:24643|issue:29870", + "issue:44568|issue:46491", + "issue:46463|issue:46899", + "issue:44821|issue:46821", + "issue:42796|issue:43906", + "issue:43452|issue:44843", + "issue:41762|issue:44936", + "issue:43720|issue:44898", + "issue:45663|issue:46846", + "issue:43653|issue:44361", + "issue:43697|issue:46829", + "issue:44913|issue:46121", + "issue:44737|issue:45588", + "issue:44756|issue:47404" + ], + "prepared_review_unit_hash": "443cf992291255607b58f0457d7c1d66b31360dde4c1df90f7653674b8a9109a", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36702, + "estimated_input_tokens": 9176, + "estimated_eval_tokens": 18608 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly heterogeneous: nearly all pairs are different bugs in unrelated models/code paths. The only strong duplicate-looking pair is the v5 tied-weights regression in translation models. No PRs are present.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 45005 is the broadest and most representative report for the only clear overlap in this set: tied-weight handling/serialization regressions in v5.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45005 is the best global anchor because it captures the underlying tied-weights regression at a broader level than the narrower device-dependent serialization report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Both involve 4-bit/quantization, but one is a torch.finfo dtype cast bug and the other is a BitsAndBytes skip-module exclusion bug; different failure modes and code paths." + }, + { + "left": "issue:43756", + "right": "issue:46399", + "accept": false, + "reason": "Different model families and defects: Smollm3 RoPE-layer dropping vs Qwen3.5 kernel registration failure." + }, + { + "left": "issue:46463", + "right": "issue:47276", + "accept": false, + "reason": "One is T5 decoder-input validation messaging; the other is a synced_gpus warning in continuous batching. Not the same bug." + }, + { + "left": "issue:44737", + "right": "issue:46463", + "accept": false, + "reason": "XLNet CPU device placement bug and T5 decoder validation error are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Different problems: Qwen3 MoE weight conversion/loading with accelerate+deepspeed vs Smollm3 RoPE layer selection." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "PEFT version bump is a dependency/versioning issue; auto_mappings referencing removed configs is a code-maintenance breakage." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and strict granite_speech config loading are different config-handling bugs." + }, + { + "left": "issue:43257", + "right": "issue:47405", + "accept": false, + "reason": "Both touch loading/conversion, but one is a Qwen3 MoE weight conversion issue and the other is a generic silent fallback when using the wrong Auto class." + }, + { + "left": "issue:43742", + "right": "issue:47332", + "accept": false, + "reason": "MobileLLM loading KeyError vs disk-offloaded save failure are different offloading/loading paths." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Both are dtype-related, but the affected models and symptoms differ: BLIP2 stays float32 vs dtype mismatches during bfloat16 inference in other wrappers." + }, + { + "left": "issue:43540", + "right": "issue:46693", + "accept": false, + "reason": "Video input validation in Qwen3OmniMoe is unrelated to the FA2 long-context performance regression." + }, + { + "left": "issue:43906", + "right": "issue:46693", + "accept": false, + "reason": "An isolated reproduction of another issue is not the same as the FA2 performance regression described here." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": true, + "reason": "Both describe the same v5 tied-weights regression: inconsistent lm_head/tied-weight serialization depending on model/device, especially for translation models." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "gguf_file breaks in tokenizer/model loading generally, while cjvt/sleng-bert appears to be a separate model-loading failure with no clear shared root cause." + }, + { + "left": "issue:24643", + "right": "issue:29870", + "accept": false, + "reason": "DeepSpeed training 'weight must be 2-D' and Bart generate decoder_attention_mask unbound local error are different failures." + }, + { + "left": "issue:44568", + "right": "issue:46491", + "accept": false, + "reason": "Different tokenizer regressions: missing BOS/EOS addition vs losing one leading space on round-trip." + }, + { + "left": "issue:46463", + "right": "issue:46899", + "accept": false, + "reason": "T5 decoder validation messaging and Gemma4 multimodal 4-bit dtype casts are unrelated." + }, + { + "left": "issue:44821", + "right": "issue:46821", + "accept": false, + "reason": "AutoImageProcessor URL loading and TimesFM window_size AttributeError are different APIs and bugs." + }, + { + "left": "issue:42796", + "right": "issue:43906", + "accept": false, + "reason": "Ministral tokenizer BPE-marker handling does not match an isolated reproduction of another issue." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "gguf_file loading breakage and unconditional model_info() in mistral regex patch are different code paths." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Gemma3 ZeRO-3 loading index error and trainer.evaluate() failing after train are unrelated trainer/loading issues." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "BitNet packed-weight unpacking during accelerate load is unrelated to Perceiver resolution/interpolation failure." + }, + { + "left": "issue:45663", + "right": "issue:46846", + "accept": false, + "reason": "Gemma4 FSDP2 shared_kv_states bug and Qwen3.5 TP plan omissions are different parallelism issues." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and MLukeTokenizer AttributeError are separate tokenizer defects." + }, + { + "left": "issue:43697", + "right": "issue:46829", + "accept": false, + "reason": "RTDetrV2 output drift in v5 and VideoPrism hidden_states wiring are different model-output bugs." + }, + { + "left": "issue:44913", + "right": "issue:46121", + "accept": false, + "reason": "GPTNeoXConfig rotary_pct reload regression and JSON-list TypeError in rope validation are separate config/validation issues." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet device placement bug and flash_attention crash on s_aux=None do not share the same underlying problem." + }, + { + "left": "issue:44756", + "right": "issue:47404", + "accept": false, + "reason": "Both mention OOM during loading, but the causes differ: mmap on Strix Halo vs unbudgeted contiguous() transients in conversion-op loading." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded in the packet and appropriately conservative overall. The only clearly duplicate-looking pair is issue:44466 \u2194 issue:45005, and the broader anchor choice for 45005 is well-supported by the titles. No PRs are present." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 27, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:33666", + "issue:36010", + "issue:38175", + "issue:40444", + "issue:43295", + "issue:43525", + "issue:43550", + "issue:43572", + "issue:43643", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43749", + "issue:43761", + "issue:43927", + "issue:44361", + "issue:44442", + "issue:44464", + "issue:44479", + "issue:44493", + "issue:44625", + "issue:44743", + "issue:44756", + "issue:44811", + "issue:44843", + "issue:44857", + "issue:44991", + "issue:45216", + "issue:45356", + "issue:45357", + "issue:45372", + "issue:45542", + "issue:45701", + "issue:45704", + "issue:45920", + "issue:46050", + "issue:46170", + "issue:46178", + "issue:46701", + "issue:46832", + "issue:46846", + "issue:46932", + "issue:47000", + "issue:47013", + "issue:47332", + "issue:47404" + ], + "soft_pairs": [ + "issue:44493|issue:46050", + "issue:44361|issue:44442", + "issue:38175|issue:43295", + "issue:46846|issue:47404", + "issue:45357|issue:45920", + "issue:44756|issue:47332", + "issue:43697|issue:45216", + "issue:40444|issue:44811", + "issue:43688|issue:43749", + "issue:44843|issue:44991", + "issue:44857|issue:47000", + "issue:45372|issue:46170", + "issue:43572|issue:43643", + "issue:43761|issue:47013", + "issue:43688|issue:45701", + "issue:45542|issue:46178", + "issue:43646|issue:43749", + "issue:43653|issue:43927", + "issue:43525|issue:46701", + "issue:33666|issue:36010", + "issue:43688|issue:45704", + "issue:44464|issue:44743", + "issue:44479|issue:44625", + "issue:43550|issue:44361", + "issue:43295|issue:45356", + "issue:46832|issue:46932" + ], + "prepared_review_unit_hash": "2f0061c3aae1f0abcd3f74f0f72957acca8e1843ef6052f3848c83129010a3a7", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 35146, + "estimated_input_tokens": 8787, + "estimated_eval_tokens": 17830 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose collection of unrelated Transformers issues, mostly centered on regressions after v5.x across tokenization, config loading, multimodal processors, and model-specific runtime bugs. None of the soft pairs looks like the same underlying bug/change, so all should stay separate.", + "confidence": 0.93, + "canonical_issue_reason": "Issue #45920 is the best representative anchor: it\u2019s a broad, concrete tokenizer regression affecting multiple models, matching the cluster\u2019s recurring theme of v5-era loading/tokenization breakage.", + "canonical_pr_reason": null, + "best_issue_reason": "#45920 is the strongest standalone issue in the set because it is clearly described, widely scoped, and central to the shared regression theme, making it the best canonical issue if one must be chosen.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44493", + "right": "issue:46050", + "accept": false, + "reason": "Different failures and code paths: a position-id key warning across many models vs a quantized model save/load corruption bug." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "Both are tokenizer-related, but they involve different tokenizer classes and distinct loading/attribute errors, not the same bug." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Unrelated symptoms: zero probabilities in SigLIP2 inference vs a regression in processor.tokenizer / image-to-tokenizer handling." + }, + { + "left": "issue:46846", + "right": "issue:47404", + "accept": false, + "reason": "Both mention OOM, but one is tensor-parallel plan sizing and the other is device_map loading transients; different mechanisms and fixes." + }, + { + "left": "issue:45357", + "right": "issue:45920", + "accept": false, + "reason": "Different layers of the stack: Qwen3.5 save_pretrained visual-key serialization vs a tokenizer ID regression across multiple models." + }, + { + "left": "issue:44756", + "right": "issue:47332", + "accept": false, + "reason": "Different offloading problems: mmap avoidance on Strix Halo to prevent OOM vs inability to save a disk-offloaded Qwen3-VL checkpoint." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "Separate model-specific regressions: RT-DETR output differences vs Qwen3.5 checkpoint save_pretrained corruption." + }, + { + "left": "issue:40444", + "right": "issue:44811", + "accept": false, + "reason": "One is finetuning with multiple images per prompt; the other is Whisper batch_decode skip_special_tokens handling. Not the same bug." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Auxiliary-loss normalization in MoE models is unrelated to FSDP CPU RAM efficient loading." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer load regressions, but one is an offline hub call in _patch_mistral_regex and the other is a specific model tokenizer load failure; no clear same code path." + }, + { + "left": "issue:44857", + "right": "issue:47000", + "accept": false, + "reason": "Both concern detection losses, but one crashes under float16 AMP/CUDA and the other has an infeasible cost matrix in RTDetrLoss; different bugs." + }, + { + "left": "issue:45372", + "right": "issue:46170", + "accept": false, + "reason": "Completely different import issues: mistral_common ReasoningEffort for Gemma 4 vs Windows/ROCm torch.distributed dependency." + }, + { + "left": "issue:43572", + "right": "issue:43643", + "accept": false, + "reason": "Both are about missing config fields, but one is a specific StableLm pad token field after v5 update and the other is a trust_remote_code AutoConfig field-loss issue." + }, + { + "left": "issue:43761", + "right": "issue:47013", + "accept": false, + "reason": "Different model families and mechanics: CLIPVision hidden_states regression vs xLSTM tuple-unpacking crash with return_last_states=False." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Auxiliary-loss normalization is not the same as a broad tokenization behavior change report." + }, + { + "left": "issue:45542", + "right": "issue:46178", + "accept": false, + "reason": "Both are backend-related, but tensorboard/TensorFlow backend detection is separate from the tensorflow_text backend mapping error." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "General custom-model initialization breakage is unrelated to FSDP CPU RAM efficient loading." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and DiaConfig custom token ID persistence are different classes of bugs." + }, + { + "left": "issue:43525", + "right": "issue:46701", + "accept": false, + "reason": "A missing pad_token_id on Llama4Config is unrelated to the DiffusionGemma image-text-to-text pipeline failure." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU Qwen2-VL training and a GenerationMixin import error are unrelated issues." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "MoE auxiliary-loss normalization is unrelated to T5 silently using apex.FusedRMSNorm and leaking memory." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Both involve generation/state, but one is compiled-forward chunked generation inconsistency and the other is xLSTM recurrent-state reset logic; not the same fix." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "A multimodal video-input regression in several Qwen models is distinct from Qwen3.5 num_labels propagation." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "torch.compile/SDPA on Bamba-9B-v2 is unrelated to MLukeTokenizer task-time AttributeError." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Processor/tokenizer API regression with images is unrelated to Kimi-K2.5 codec handling and mistral regex warning behavior." + }, + { + "left": "issue:46832", + "right": "issue:46932", + "accept": false, + "reason": "RT-DETR forward shape/type error is unrelated to Gemma4 quantized vision/audio embedder dtype casting." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 28, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28370", + "issue:32090", + "issue:33666", + "issue:34567", + "issue:36331", + "issue:41720", + "issue:42898", + "issue:43278", + "issue:43452", + "issue:43504", + "issue:43525", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43611", + "issue:43643", + "issue:43653", + "issue:43688", + "issue:43720", + "issue:43756", + "issue:43761", + "issue:43782", + "issue:43927", + "issue:44361", + "issue:44464", + "issue:44466", + "issue:44554", + "issue:44561", + "issue:44568", + "issue:44610", + "issue:44625", + "issue:44734", + "issue:44964", + "issue:44977", + "issue:45405", + "issue:45440", + "issue:45823", + "issue:46050", + "issue:46402", + "issue:46491", + "issue:46612", + "issue:46693", + "issue:46829", + "issue:46846", + "issue:47254", + "issue:47276" + ], + "soft_pairs": [ + "issue:43756|issue:46402", + "issue:43761|issue:46829", + "issue:43653|issue:44568", + "issue:41720|issue:44464", + "issue:33666|issue:36331", + "issue:43452|issue:43577", + "issue:28370|issue:34567", + "issue:44466|issue:46050", + "issue:44977|issue:46693", + "issue:43504|issue:43550", + "issue:44464|issue:44977", + "issue:32090|issue:33666", + "issue:42898|issue:46491", + "issue:45405|issue:47254", + "issue:43606|issue:43927", + "issue:43720|issue:43927", + "issue:43782|issue:46693", + "issue:43504|issue:43720", + "issue:43278|issue:43688", + "issue:43643|issue:44561", + "issue:44625|issue:47276", + "issue:43720|issue:44361", + "issue:43504|issue:44610", + "issue:43688|issue:45440", + "issue:44734|issue:46612", + "issue:43525|issue:43572", + "issue:43577|issue:44554", + "issue:44964|issue:45405", + "issue:45823|issue:46050", + "issue:43611|issue:46846" + ], + "prepared_review_unit_hash": "401ab4d89f30dc340aee34b7e3578c004c429ad5851ab7fd947c15bd5d82c3f7", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36299, + "estimated_input_tokens": 9075, + "estimated_eval_tokens": 18406 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer, config, loading, training, generation, and serving bugs. None of the soft pairs looks like the same underlying fix, so no duplicate merge is justified.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits the cluster because the items span unrelated regressions and model families rather than one shared bug.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a good global representative; the set is too broad and the reported failures are unrelated in codepath and fix.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43756", + "right": "issue:46402", + "accept": false, + "reason": "Different models and different failures: RoPE layer dropping in Smollm3 versus text encoder weights not loading in Sam3Model." + }, + { + "left": "issue:43761", + "right": "issue:46829", + "accept": false, + "reason": "Both mention hidden states, but they are separate model-specific output bugs with different APIs and symptoms." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer special-token registration for BigBird is not the same bug as BOS/EOS insertion behavior for mdeberta-v3." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "CUDA assert during Qwen3 device mapping is unrelated to chunked generation inconsistencies under compiled forward." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Multi-GPU training on Qwen2-VL is not the same as a Trainer API kwarg change breaking custom compute_loss." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "gguf_file loading breakage and BLIP2 dtype persistence are different loading paths with unrelated root causes." + }, + { + "left": "issue:28370", + "right": "issue:34567", + "accept": false, + "reason": "Missing vocab_file in custom SentencePiece models is unrelated to TrainerState.num_input_tokens_seen not updating." + }, + { + "left": "issue:44466", + "right": "issue:46050", + "accept": false, + "reason": "Serialization of tied lm_head weights is a different bug from quantized sinq save/load corruption." + }, + { + "left": "issue:44977", + "right": "issue:46693", + "accept": false, + "reason": "Flash-attention generation correctness on Qwen3.5 is not the same underlying issue as a long-context FA2 performance regression." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Legacy-field model loading in Beit is unrelated to a torch.compile + SDPA failure in Bamba." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Chunked generation with compiled forward and flash-attention generation failures are distinct codepaths and fixes." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "A NoneType GPU broadcast error in Trainer is not the same as Qwen2-VL multi-GPU training behavior." + }, + { + "left": "issue:42898", + "right": "issue:46491", + "accept": false, + "reason": "Global clean_up_tokenization_spaces behavior changes are unrelated to CodeLlama's leading-space round-trip regression." + }, + { + "left": "issue:45405", + "right": "issue:47254", + "accept": false, + "reason": "A release/version pin issue for PEFT is not the same as a runtime CheckpointError with ZeRO-3 and gradient checkpointing." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch in Bark is unrelated to DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "Packed-weight unpacking during accelerate loading is a different failure mode from config field loss after serialization." + }, + { + "left": "issue:43782", + "right": "issue:46693", + "accept": false, + "reason": "Qwen3VL weight_only loading error is unrelated to the FA2 long-context performance regression." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "A legacy field blocking Beit preset loading is not the same as BitNet packed-weight handling in accelerate." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Embedding dtype drift between train/eval and auxiliary loss normalization are separate numerical issues." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both involve trust_remote_code, but one is missing fields in AutoConfig output while the other is an API removal breaking remote models." + }, + { + "left": "issue:44625", + "right": "issue:47276", + "accept": false, + "reason": "Qwen3.5 num_labels propagation is unrelated to a misleading synced_gpus warning message." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "BitNet packed-weight loading and MLukeTokenizer AttributeError affect different subsystems and codepaths." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Legacy-field loading in Beit is unrelated to OmDet-Turbo's processor/model input-size mismatch." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Auxiliary loss normalization is a different bug from DeepseekV3 implementation divergence." + }, + { + "left": "issue:44734", + "right": "issue:46612", + "accept": false, + "reason": "KV-cache continuation indexing in serve is not the same as beam-search cache reorder being skipped for specific models." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config and missing pad_token_idx in StableLmConfig are separate config-field regressions." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "BLIP2 dtype persistence is unrelated to an MPS attention correctness issue when value head dim differs from query." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Phi-4 multimodal loading failure and unreleased PEFT version pin are unrelated." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Gemma4 PLE device mismatch during forward is a different bug from quantized sinq save/load breakage." + }, + { + "left": "issue:43611", + "right": "issue:46846", + "accept": false, + "reason": "base_model_prefix load regression and missing linear_attn entries in base_model_tp_plan are separate model-loading and tensor-parallel planning issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 29, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28218", + "issue:29925", + "issue:33357", + "issue:39692", + "issue:42796", + "issue:43257", + "issue:43278", + "issue:43295", + "issue:43334", + "issue:43493", + "issue:43504", + "issue:43531", + "issue:43572", + "issue:43606", + "issue:43646", + "issue:43701", + "issue:43825", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44291", + "issue:44361", + "issue:44466", + "issue:44484", + "issue:44488", + "issue:44514", + "issue:44554", + "issue:44568", + "issue:44589", + "issue:44704", + "issue:44843", + "issue:44977", + "issue:45290", + "issue:45405", + "issue:45491", + "issue:45520", + "issue:45759", + "issue:45812", + "issue:46002", + "issue:46192", + "issue:46209", + "issue:46365", + "issue:46489", + "issue:47332", + "issue:47404", + "issue:47436" + ], + "soft_pairs": [ + "issue:43334|issue:45759", + "issue:28218|issue:33357", + "issue:44484|issue:44704", + "issue:44514|issue:45290", + "issue:39692|issue:43295", + "issue:43334|issue:43572", + "issue:45405|issue:47332", + "issue:43994|issue:46192", + "issue:44488|issue:44843", + "issue:44977|issue:46209", + "issue:43257|issue:47404", + "issue:43493|issue:43825", + "issue:42796|issue:44977", + "issue:43531|issue:47436", + "issue:43927|issue:44361", + "issue:43646|issue:44466", + "issue:43572|issue:44291", + "issue:45491|issue:46002", + "issue:43278|issue:43701", + "issue:44554|issue:46365", + "issue:28218|issue:29925", + "issue:44568|issue:45812", + "issue:44568|issue:46489", + "issue:43873|issue:47404", + "issue:43504|issue:43606", + "issue:44589|issue:45520", + "issue:43278|issue:43844", + "issue:43493|issue:43844" + ], + "prepared_review_unit_hash": "496cd4db430b1a80ba16274a700d84b392058db5ff6b5908630976c0b09b4547", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35713, + "estimated_input_tokens": 8929, + "estimated_eval_tokens": 18114 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The soft links are mostly false positives across unrelated transformer bugs. I would anchor the cluster on the broad, high-activity offloading/quantization report, and I do not see any pair here that is a true duplicate or same-code-path change.", + "confidence": 0.72, + "canonical_issue_reason": "Issue 43873 is the most active and broadly phrased report in the set (highest discussion/inbound activity, still open), so it is the best anchor for triage.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43873 is the strongest standalone bug report here: high signal, broad impact, and enough activity to serve as the representative issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Different failures in Qwen multimodal config handling: missing pad_token_id vs unwrapping text_config in from_config." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer spacing after added tokens is unrelated to a Mac bus error in a CLIP model." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "save_pretrained shard-size question and AutoProcessor kwargs passthrough are different issues." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both hit apply_chat_template, but the triggers and affected inputs differ materially." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Docs example mistakes are not the same bug as a processor.tokenizer/images regression." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Different model families and different missing config fields." + }, + { + "left": "issue:45405", + "right": "issue:47332", + "accept": false, + "reason": "PEFT version bump issue and disk-offload save failure are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 output quality regression is not the same as silently ignored return_tensors kwargs." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "A model-loading failure and an offline-mode network call in tokenizer patching are distinct code paths." + }, + { + "left": "issue:44977", + "right": "issue:46209", + "accept": false, + "reason": "Flash-attention generation problems are not the same as FP8/B200 inference failures." + }, + { + "left": "issue:43257", + "right": "issue:47404", + "accept": false, + "reason": "Qwen3 MoE weight conversion under accelerate+deepspeed is different from device_map transient OOMs." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "Model implementation discrepancy and pipeline error-message wording are unrelated." + }, + { + "left": "issue:42796", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer BPE-marker handling does not match a Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:43531", + "right": "issue:47436", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is not the same as a missing max_position_embeddings attribute." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "save/load token-id loss and tokenizer AttributeError on tasks are different defects." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model initialization breakage is distinct from tied-weight serialization varying by device." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "A missing pad_token_idx field and an init_empty_weights argument mismatch are unrelated." + }, + { + "left": "issue:45491", + "right": "issue:46002", + "accept": false, + "reason": "Gemma3 sliding-window NaNs and classification-head NaNs after checkout/init are different symptoms with different causes." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Embedding dtype drift is not the same as resume_from_checkpoint key mismatch." + }, + { + "left": "issue:44554", + "right": "issue:46365", + "accept": false, + "reason": "MPS attention correctness and Trainer import conflict with datasets are unrelated." + }, + { + "left": "issue:28218", + "right": "issue:29925", + "accept": false, + "reason": "Extra space after added token and SigLIP fast-tokenizer enforcement do not indicate the same bug." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "mDeBERTa special-token regression and Granite token-id regression are separate tokenizer bugs." + }, + { + "left": "issue:44568", + "right": "issue:46489", + "accept": false, + "reason": "Different models, different symptoms, and no concrete shared code path are shown." + }, + { + "left": "issue:43873", + "right": "issue:47404", + "accept": false, + "reason": "Quantization/offloading not working as expected is not the same as loader transient OOM from conversion ops." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "Legacy-field model loading failure and CPU offload device mismatch are different issues." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Float8 storage lookup failure and flash_attn import KeyError are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Embedding dtype drift does not match a ZeRO-3 gradient growth bug." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 HF-vs-JAX discrepancy is unrelated to gradient instability under HfDeepSpeedConfig." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 30, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:15354", + "issue:28218", + "issue:34567", + "issue:37428", + "issue:38175", + "issue:41720", + "issue:42490", + "issue:42947", + "issue:43381", + "issue:43493", + "issue:43504", + "issue:43572", + "issue:43606", + "issue:43653", + "issue:43696", + "issue:43701", + "issue:43824", + "issue:43844", + "issue:43873", + "issue:43957", + "issue:43994", + "issue:44112", + "issue:44423", + "issue:44464", + "issue:44493", + "issue:44589", + "issue:44734", + "issue:44849", + "issue:44964", + "issue:45072", + "issue:45084", + "issue:45200", + "issue:45229", + "issue:45245", + "issue:45357", + "issue:45520", + "issue:45584", + "issue:45588", + "issue:45823", + "issue:46050", + "issue:46209", + "issue:46285", + "issue:46710", + "issue:46821", + "issue:46832", + "issue:47013", + "issue:47254", + "issue:47436" + ], + "soft_pairs": [ + "issue:46821|issue:46832", + "issue:43824|issue:44589", + "issue:46209|issue:46710", + "issue:42490|issue:43696", + "issue:43504|issue:43653", + "issue:43606|issue:44112", + "issue:43994|issue:44493", + "issue:44493|issue:46285", + "issue:43572|issue:47436", + "issue:45072|issue:46832", + "issue:43873|issue:46050", + "issue:44464|issue:45584", + "issue:15354|issue:38175", + "issue:45245|issue:46710", + "issue:47013|issue:47254", + "issue:44849|issue:45357", + "issue:45084|issue:45520", + "issue:42947|issue:43381", + "issue:45229|issue:45823", + "issue:44423|issue:44734", + "issue:44849|issue:46209", + "issue:43844|issue:44964", + "issue:28218|issue:34567", + "issue:45588|issue:46050", + "issue:45200|issue:45229", + "issue:37428|issue:43957", + "issue:43493|issue:43701", + "issue:41720|issue:46710" + ], + "prepared_review_unit_hash": "7a02a1f3c927689686e84c948dc0131a687e9049e9bf0cebfa740fa6941a31b0", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36209, + "estimated_input_tokens": 9053, + "estimated_eval_tokens": 18362 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The items are a heterogeneous set of unrelated issues, not a duplicate cluster. None of the soft pairs describe the same underlying bug or change closely enough to merge/close as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": "If a representative issue must be chosen, issue #43873 is the strongest single candidate because it has the richest discussion and references, but the overall set is too diverse to justify a true canonical duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #43873 is the best representative of the set only in the sense of being the most substantial/actively discussed issue; it is not a duplicate anchor for the other items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46821", + "right": "issue:46832", + "accept": false, + "reason": "Both are runtime errors in model code, but they affect different models, different inputs, and different failure paths." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "An import error for a missing Qwen class and a storage-object CUDA type error are unrelated bugs." + }, + { + "left": "issue:46209", + "right": "issue:46710", + "accept": false, + "reason": "FP8 inference on B200 and an accuracy regression after upgrade are different problems with different symptoms and likely causes." + }, + { + "left": "issue:42490", + "right": "issue:43696", + "accept": false, + "reason": "A local-path import warning and a GPU out-of-memory error are not the same underlying issue." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "Loading a BEiT preset with a legacy field and BigBird tokenizer special-token registration are unrelated model/tokenizer bugs." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "A real CPU-offload device mismatch in Bark is not the same as a stale CI test failure in GraniteSpeech." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 producing nonsense outputs and a widespread position-id key warning point to different code paths and failure modes." + }, + { + "left": "issue:44493", + "right": "issue:46285", + "accept": false, + "reason": "A position-id key issue across models is not the same as a CLIP text-model structure inconsistency." + }, + { + "left": "issue:43572", + "right": "issue:47436", + "accept": false, + "reason": "Both mention missing config attributes, but they concern different models and different missing fields." + }, + { + "left": "issue:45072", + "right": "issue:46832", + "accept": false, + "reason": "A dtype mismatch in inference is unrelated to an RT-DETR feature-level indexing/type error." + }, + { + "left": "issue:43873", + "right": "issue:46050", + "accept": false, + "reason": "Quantization/offloading behavior and save/load corruption of sinq-quantized models are distinct bugs." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency is unrelated to empty-transcription generation after token alignment." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "A torch.jit.script GeneratorExp export failure and unexpected zero probabilities in SigLIP2 are different issues." + }, + { + "left": "issue:45245", + "right": "issue:46710", + "accept": false, + "reason": "Category-count overflow and an accuracy regression after version upgrade are not the same bug." + }, + { + "left": "issue:47013", + "right": "issue:47254", + "accept": false, + "reason": "An xLSTM train-mode unpacking crash and a PEFT/DeepSpeed checkpoint error are different code-path failures." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Qwen3.5 output_hidden_states behavior and incorrect visual encoder keys on save_pretrained are unrelated regressions." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "A template-node compilation TypeError and a Python 3.13 flash_attn import KeyError are unrelated." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Gradient checkpointing ineffective in one case and disallowed in eval mode in another are different behaviors, not duplicates." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Multi-GPU CUDA OOM and device-mismatch with auto device mapping are different problems." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Continuous-batching multimodal crash and KV-cache continuation indexing bug are separate serving failures." + }, + { + "left": "issue:44849", + "right": "issue:46209", + "accept": false, + "reason": "A Qwen3.5 hidden-states bug and an FP8 inference failure have no shared underlying cause." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "A DeepSpeed ZeRO-3 gradient anomaly and a model-loading failure for Phi-4 multimodal are unrelated." + }, + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer spacing after added tokens and TrainerState token accounting are unrelated." + }, + { + "left": "issue:45588", + "right": "issue:46050", + "accept": false, + "reason": "A flash-attention crash on sink-less models and quantized save/load breakage are different bugs." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Missing mm_token_type_ids defaulting and multi-GPU OOM are unrelated issues." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "A missing flash-attention import symbol and meta-device model loading failures are different regressions." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "A SigLIP2 implementation discrepancy and a resume_from_checkpoint key mismatch do not describe the same bug." + }, + { + "left": "issue:41720", + "right": "issue:46710", + "accept": false, + "reason": "Auto device mapping causing cudaErrorAssert and an accuracy regression are distinct failure modes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 31, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:27957", + "issue:28370", + "issue:29942", + "issue:30333", + "issue:34689", + "issue:37428", + "issue:42915", + "issue:43065", + "issue:43295", + "issue:43334", + "issue:43404", + "issue:43450", + "issue:43493", + "issue:43525", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43575", + "issue:43824", + "issue:43844", + "issue:43873", + "issue:43937", + "issue:44186", + "issue:44554", + "issue:44610", + "issue:44849", + "issue:44869", + "issue:44977", + "issue:45072", + "issue:45198", + "issue:45200", + "issue:45406", + "issue:45446", + "issue:45520", + "issue:45588", + "issue:45759", + "issue:45823", + "issue:46121", + "issue:46178", + "issue:46192", + "issue:46369", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46821", + "issue:47145", + "issue:47332", + "issue:47404" + ], + "soft_pairs": [ + "issue:44849|issue:47145", + "issue:47332|issue:47404", + "issue:43295|issue:43450", + "issue:43937|issue:46701", + "issue:43065|issue:46369", + "issue:44977|issue:46701", + "issue:44554|issue:45446", + "issue:45072|issue:45198", + "issue:45588|issue:45759", + "issue:43404|issue:46710", + "issue:43295|issue:43873", + "issue:44869|issue:46752", + "issue:43572|issue:46821", + "issue:29942|issue:30333", + "issue:28370|issue:34689", + "issue:43540|issue:45200", + "issue:43295|issue:44610", + "issue:45823|issue:46192", + "issue:45406|issue:46121", + "issue:43450|issue:43575", + "issue:44186|issue:46192", + "issue:37428|issue:43824", + "issue:43550|issue:43844", + "issue:45200|issue:45823", + "issue:45520|issue:46178", + "issue:43334|issue:43525", + "issue:27957|issue:28370", + "issue:42915|issue:43824", + "issue:43295|issue:43493" + ], + "prepared_review_unit_hash": "87d0ce9f012e9f0a68350d54f686d5182b81a332a840bf6adc8b7dfd4f1f343c", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36408, + "estimated_input_tokens": 9102, + "estimated_eval_tokens": 18460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the issues span unrelated bugs in model loading, attention backends, tokenizers/processors, offloading, compilation, and training/runtime behavior. The soft edges are mostly keyword overlap (e.g. FlashAttention, padding, offloading) rather than the same underlying defect, so they should be rejected.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue exists; these reports describe different bugs in different code paths and models, so there is no true duplicate center for the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43873 is the strongest standalone representative only because it has the most discussion and inbound references, but it is still not a duplicate anchor for the rest of the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44849", + "right": "issue:47145", + "accept": false, + "reason": "Different bugs: Qwen3.5 hidden-states behavior vs FP8 compressed-tensors checkpoints missing weight_scale application." + }, + { + "left": "issue:47332", + "right": "issue:47404", + "accept": false, + "reason": "Both involve loading/offloading, but one is a save failure for disk-offloaded Qwen3-VL and the other is an OOM from conversion-op transients during device_map loading." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor.tokenizer regression is unrelated to batched video shape handling." + }, + { + "left": "issue:43937", + "right": "issue:46701", + "accept": false, + "reason": "GLM-5 GenerationConfig validation and DiffusionGemma image-text pipeline failure are different code paths." + }, + { + "left": "issue:43065", + "right": "issue:46369", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as a memory overflow in sam3-litetext." + }, + { + "left": "issue:44977", + "right": "issue:46701", + "accept": false, + "reason": "Flash-attention generation issue for Qwen3.5 is unrelated to the DiffusionGemma pipeline bug." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "MPS attention correctness and a PyTorch version-check import bug are separate problems." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "Bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel are unrelated to Wav2Vec2 save/tokenization failures." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "A flash_attention helper crash on sink-less models is different from not unwrapping text_config in composite Qwen configs." + }, + { + "left": "issue:43404", + "right": "issue:46710", + "accept": false, + "reason": "Untied lm_head weights in Mistral3ForConditionalGeneration is not the same as a DeepSeek accuracy regression after upgrade." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Processor/tokenizer regression and quantization offloading issues are unrelated." + }, + { + "left": "issue:44869", + "right": "issue:46752", + "accept": false, + "reason": "Whisper timestamp decode crash and empty-chat-template IndexError are different text-processing bugs." + }, + { + "left": "issue:43572", + "right": "issue:46821", + "accept": false, + "reason": "StableLm pad_token_idx regression and TimesFM window_size AttributeError are separate config issues." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 test failures do not match an MLFlow reporting issue after failed training jobs." + }, + { + "left": "issue:28370", + "right": "issue:34689", + "accept": false, + "reason": "Missing vocab_file for custom SentencePiece models is unrelated to Llama 3.2 Vision loading breakage." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError and Gemma 4 mm_token_type_ids defaulting are different multimodal bugs." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "A processor.tokenizer regression is not the same as OmDet-Turbo producing the wrong image size." + }, + { + "left": "issue:45823", + "right": "issue:46192", + "accept": false, + "reason": "Gemma4 PLE device mismatch and return_tensors being ignored are unrelated behaviors." + }, + { + "left": "issue:45406", + "right": "issue:46121", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and convert_rope_params_to_dict TypeError are different failures." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Batched video shape errors are unrelated to Qwen2-57B-A14B-Instruct tensor-parallel OOM." + }, + { + "left": "issue:44186", + "right": "issue:46192", + "accept": false, + "reason": "LayoutLMv2Tokenizer crashes on NER/padding are not the same as return_tensors being ignored with text_kwargs." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "These are both import errors, but for different missing symbols in different modules and model families." + }, + { + "left": "issue:43550", + "right": "issue:43844", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to ZeRO-3 gradient growth." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Gemma 4 token-type defaults and Gemma4 PLE device placement are separate problems." + }, + { + "left": "issue:45520", + "right": "issue:46178", + "accept": false, + "reason": "flash_attn KeyError on Python 3.13 is unrelated to BACKENDS_MAPPING rejecting tensorflow_text." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Both mention pad_token_id, but they affect different config classes/models and are not the same concrete bug." + }, + { + "left": "issue:27957", + "right": "issue:28370", + "accept": false, + "reason": "XLMRoberta FlashAttention 2 support and missing vocab_file for custom SentencePiece models are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "FineGrainedFP8Config failure in Qwen3Moe is unrelated to the Qwen2_5_VL import error." + }, + { + "left": "issue:43295", + "right": "issue:43493", + "accept": false, + "reason": "Processor/tokenizer regression and SigLIP2 implementation discrepancy are distinct bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 32, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28218", + "issue:29870", + "issue:29942", + "issue:35141", + "issue:37428", + "issue:39692", + "issue:42490", + "issue:43262", + "issue:43278", + "issue:43335", + "issue:43425", + "issue:43450", + "issue:43452", + "issue:43493", + "issue:43550", + "issue:43611", + "issue:43653", + "issue:43819", + "issue:43824", + "issue:43856", + "issue:43950", + "issue:43975", + "issue:44112", + "issue:44442", + "issue:44466", + "issue:44479", + "issue:44521", + "issue:44561", + "issue:44610", + "issue:44737", + "issue:44857", + "issue:44898", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45290", + "issue:45325", + "issue:45381", + "issue:45412", + "issue:45520", + "issue:46192", + "issue:46209", + "issue:46234", + "issue:46285", + "issue:46512", + "issue:46894", + "issue:46899", + "issue:47405" + ], + "soft_pairs": [ + "issue:42490|issue:43493", + "issue:29870|issue:29942", + "issue:39692|issue:46192", + "issue:43824|issue:43975", + "issue:43278|issue:43856", + "issue:43262|issue:43450", + "issue:43975|issue:45520", + "issue:28218|issue:35141", + "issue:44521|issue:45290", + "issue:43653|issue:44112", + "issue:43950|issue:46285", + "issue:44737|issue:46894", + "issue:43452|issue:46512", + "issue:44442|issue:45198", + "issue:44479|issue:46192", + "issue:43819|issue:44112", + "issue:44466|issue:45127", + "issue:43335|issue:43550", + "issue:44561|issue:46234", + "issue:45200|issue:45325", + "issue:43611|issue:47405", + "issue:45381|issue:46209", + "issue:44898|issue:45412", + "issue:45200|issue:46899", + "issue:44610|issue:44857", + "issue:37428|issue:43425" + ], + "prepared_review_unit_hash": "1608473db0e1fa76e9529b3391dab5486292ae23c3c54f6d81afbd72fff67edc", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 36014, + "estimated_input_tokens": 9004, + "estimated_eval_tokens": 18264 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The soft-pair suggestions are mostly false positives: the issues touch different models, APIs, or failure modes, and none of the PR-style merges are warranted here. I\u2019d use issue #44521 as the best representative issue because it is the most active and broadly referenced item in the set, but there is no true duplicate cluster.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #44521 is the strongest representative item by activity and inbound references, and it is an active, user-facing bug with broad discussion.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #44521 is the most actionable and central issue in the set; it has the most discussion/ref activity and a clear bug report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42490", + "right": "issue:43493", + "accept": false, + "reason": "Different problems: local-path loading warning vs. SigLIP2 implementation discrepancy." + }, + { + "left": "issue:29870", + "right": "issue:29942", + "accept": false, + "reason": "Bart generate() decoder mask bug is unrelated to Flash Attention 2 test failures." + }, + { + "left": "issue:39692", + "right": "issue:46192", + "accept": false, + "reason": "Documentation/example errors are unrelated to `return_tensors` being ignored when `text_kwargs` is passed." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Qwen2.5-VL import error and DeepSeek detokenization are different code paths and symptoms." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "Embedding dtype drift during evaluate is unrelated to Qwen3 MoE memory usage." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and batched video shape handling are separate processor bugs." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Detokenization corruption and Python 3.13 `flash_attn` import errors do not share the same underlying fix." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Extra tokenizer space after added token is unrelated to output embedding reinitialization after resizing." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both involve `apply_chat_template`, but one is zeroed assistant masks for multimodal inputs and the other is a crash on tool-call messages; different bugs." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and a stale GraniteSpeech CI test are unrelated." + }, + { + "left": "issue:43950", + "right": "issue:46285", + "accept": false, + "reason": "Non-persistent buffer corruption on `from_pretrained()` is unrelated to CLIP text model structure changes." + }, + { + "left": "issue:44737", + "right": "issue:46894", + "accept": false, + "reason": "XLNet CPU `arange` device bug and CohereASR training-loss double-shift are different model-specific issues." + }, + { + "left": "issue:43452", + "right": "issue:46512", + "accept": false, + "reason": "GGUF loading failures are unrelated to `from_config(dtype=...)` leaving `config.dtype` stale." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "FastSpeech2Conformer AutoTokenizer loading failure and Wav2Vec2 save/tokenization failure are not the same bug." + }, + { + "left": "issue:44479", + "right": "issue:46192", + "accept": false, + "reason": "Video input regression for Qwen models is unrelated to `return_tensors` handling in text kwargs." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC latent reconstruction mismatch and a GraniteSpeech CI test failure are unrelated." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Serialization inconsistency for `lm_head.weight` is a different issue from LoRA merge collapse with extended vocabulary." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug is unrelated to Bamba torch.compile/SDPA failure." + }, + { + "left": "issue:44561", + "right": "issue:46234", + "accept": false, + "reason": "Removal of `is_torch_fx_available` breaking trust_remote_code models is unrelated to missing `is_falcon_mamba_ssm_available`." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Gemma4 `mm_token_type_ids` defaults and Qwen2.5-VL rope-index scaling are different multimodal bugs." + }, + { + "left": "issue:43611", + "right": "issue:47405", + "accept": false, + "reason": "Base-model-prefix loading regression and silent wrong-Auto-class conversion are distinct loader failures." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "Qwen2.5-VL vision position IDs and FP8 inference on B200 are unrelated." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver non-default-resolution failure is unrelated to RT-DETR memory leak on deletion." + }, + { + "left": "issue:45200", + "right": "issue:46899", + "accept": false, + "reason": "Gemma4 text-only `mm_token_type_ids` defaulting and 4-bit quantization dtype-cast failures are different multimodal issues." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "OmDet-Turbo processor size mismatch and LwDetrImageLoss AMP crash are separate problems." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Missing flash-attention import symbol and Torch 2.10 incompatibility are not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 33, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28282", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:38175", + "issue:41720", + "issue:42490", + "issue:43335", + "issue:43504", + "issue:43525", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43819", + "issue:43827", + "issue:43854", + "issue:43901", + "issue:44112", + "issue:44206", + "issue:44292", + "issue:44493", + "issue:44617", + "issue:44792", + "issue:44843", + "issue:44855", + "issue:44857", + "issue:45083", + "issue:45200", + "issue:45412", + "issue:45520", + "issue:45563", + "issue:45584", + "issue:45593", + "issue:45759", + "issue:45923", + "issue:46209", + "issue:46285", + "issue:46424", + "issue:46431", + "issue:46512", + "issue:46566", + "issue:46697", + "issue:46821", + "issue:46858", + "issue:46932", + "issue:47405", + "issue:47427", + "issue:47436" + ], + "soft_pairs": [ + "issue:45200|issue:46932", + "issue:44843|issue:47427", + "issue:43819|issue:45083", + "issue:43335|issue:43653", + "issue:43606|issue:45083", + "issue:43525|issue:47436", + "issue:44857|issue:45593", + "issue:41720|issue:46209", + "issue:44792|issue:45563", + "issue:28282|issue:33357", + "issue:43335|issue:43504", + "issue:46512|issue:46821", + "issue:43550|issue:44112", + "issue:45923|issue:47405", + "issue:38175|issue:45412", + "issue:45759|issue:46285", + "issue:43827|issue:43901", + "issue:29942|issue:32090", + "issue:44206|issue:45584", + "issue:46431|issue:46566", + "issue:44855|issue:45083", + "issue:43335|issue:43606", + "issue:43335|issue:44112", + "issue:44617|issue:46697", + "issue:42490|issue:43854", + "issue:43819|issue:44493", + "issue:44292|issue:45520", + "issue:46424|issue:46858", + "issue:43335|issue:45083" + ], + "prepared_review_unit_hash": "d6987566fca0ac4aca26197a72547c19912351c4f0ba51a1bf8b70d65a424242", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36672, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster does not form a true duplicate set: the issues span unrelated models, docs, tokenizer, cache, device-mapping, and CI failures. All soft pairs should be rejected.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue stands out; the items are thematically unrelated and the soft links are only loose title similarity, not the same underlying bug.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue is suitable as a representative duplicate target because there is no shared code-path or common fix across the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45200", + "right": "issue:46932", + "accept": false, + "reason": "Different Gemma 4 problems: missing mm_token_type_ids default vs quantized vision/audio dtype casting." + }, + { + "left": "issue:44843", + "right": "issue:47427", + "accept": false, + "reason": "Offline Mistral regex patch bug vs redundant init skipping in _is_hf_initialized; unrelated code paths." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC latent decoding mismatch and qwen3_omni_moe length helper behavior are different bugs." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug and BigBirdTokenizer special-token registration are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "CPU offload device mismatch in bark-small is unrelated to qwen3_omni_moe length calculation." + }, + { + "left": "issue:43525", + "right": "issue:47436", + "accept": false, + "reason": "Different missing-config-attribute bugs on different models: pad_token_id vs max_position_embeddings." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "Different model losses: LwDetr float16 AMP crash vs D-FINE auxiliary-loss handling." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "Both are Qwen-related but the failures differ: cudaErrorAssert on auto device map vs FP8 inference on B200." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "janus image-generation test failure and paged-generate warning are not the same underlying issue." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "PyTorch missing ImportError and MacOS bus error in CLIP are entirely different failures." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "SwitchTransformers sparse-layer creation and Beit legacy-field loading are unrelated model-specific bugs." + }, + { + "left": "issue:46512", + "right": "issue:46821", + "accept": false, + "reason": "Stale config.dtype after from_config is unrelated to TimesFM window_size attribute errors." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure and GraniteSpeech stale device-override CI test are different problems." + }, + { + "left": "issue:45923", + "right": "issue:47405", + "accept": false, + "reason": "Missing supports_gradient_checkpointing flag is unrelated to silent conversion skipping on model_type mismatch." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "Zero probabilities in SigLIP2 and RT-DETR memory retention have no shared fix path." + }, + { + "left": "issue:45759", + "right": "issue:46285", + "accept": false, + "reason": "Composite Qwen from_config unwrapping and CLIP text_model structure drift are different model-API issues." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs issues, but they cover different APIs and different deprecated behaviors." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 test failures and _gpu_broadcast_one NoneType errors are unrelated trainer/runtime bugs." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Unsupported center arg in feature extractor and Whisper empty-transcription failure are different code paths." + }, + { + "left": "issue:46431", + "right": "issue:46566", + "accept": false, + "reason": "Missing chat template error and DiffusionGemma meta-device loading are unrelated loading issues." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "DebertaV2 import IndentationError and qwen3_omni_moe helper behavior are unrelated." + }, + { + "left": "issue:43335", + "right": "issue:43606", + "accept": false, + "reason": "SwitchTransformers sparse-layer config and bark-small CPU offload device mismatch do not share a fix." + }, + { + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug and GraniteSpeech stale device override test are unrelated." + }, + { + "left": "issue:44617", + "right": "issue:46697", + "accept": false, + "reason": "Sam3Video OOM and Swin2SR over-padding are unrelated issues." + }, + { + "left": "issue:42490", + "right": "issue:43854", + "accept": false, + "reason": "Trailing-slash import warning and GLM-4.7-Flash unit-test load failure are different loading problems." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "DAC.from_latents mismatch and unexpected position-id key warnings are distinct model bugs." + }, + { + "left": "issue:44292", + "right": "issue:45520", + "accept": false, + "reason": "Qwen-3-8B-NVFP4 runtime failure and flash_attn Python 3.13 import KeyError are unrelated." + }, + { + "left": "issue:46424", + "right": "issue:46858", + "accept": false, + "reason": "Cache max_cache_length behavior and device-tensor seq_length arithmetic are different cache bugs." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "SwitchTransformers sparse-layer creation and qwen3_omni_moe feature-length helper behavior are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 34, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:10000", + "issue:15354", + "issue:25251", + "issue:27957", + "issue:29127", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:42175", + "issue:42490", + "issue:43278", + "issue:43329", + "issue:43388", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43701", + "issue:43720", + "issue:43825", + "issue:44112", + "issue:44206", + "issue:44351", + "issue:44448", + "issue:44493", + "issue:44514", + "issue:44521", + "issue:44743", + "issue:44779", + "issue:44857", + "issue:45072", + "issue:45083", + "issue:45290", + "issue:45310", + "issue:45459", + "issue:45478", + "issue:45588", + "issue:45593", + "issue:45823", + "issue:45859", + "issue:46234", + "issue:46326", + "issue:47000" + ], + "soft_pairs": [ + "issue:43388|issue:43526", + "issue:45588|issue:45859", + "issue:37428|issue:44351", + "issue:43606|issue:44857", + "issue:44206|issue:45823", + "issue:39692|issue:44493", + "issue:44448|issue:44743", + "issue:10000|issue:15354", + "issue:39692|issue:43526", + "issue:45290|issue:46326", + "issue:34689|issue:35141", + "issue:44514|issue:44521", + "issue:43720|issue:44112", + "issue:43701|issue:45083", + "issue:44857|issue:45072", + "issue:29127|issue:32090", + "issue:44448|issue:44779", + "issue:32090|issue:33357", + "issue:43550|issue:45083", + "issue:38175|issue:44493", + "issue:42490|issue:43575", + "issue:37428|issue:42175", + "issue:43278|issue:43825", + "issue:27957|issue:34689", + "issue:43278|issue:43329", + "issue:45310|issue:45478", + "issue:25251|issue:32090", + "issue:29942|issue:35141", + "issue:45593|issue:47000", + "issue:45459|issue:46234" + ], + "prepared_review_unit_hash": "3b0b717e9bfac1fbe8181af91fc915c66c36bfbef3c1ede7ddb04fdd380bb30a", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36847, + "estimated_input_tokens": 9212, + "estimated_eval_tokens": 18680 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most candidate edges are not true duplicates: they share a subsystem or symptom category, but the underlying failures differ. The only clear duplicate is the repeated Qwen3.5 Moe `from_pretrained` regression report (issues 45310 and 45478).", + "confidence": 0.88, + "canonical_issue_reason": "Issue 45310 is the better canonical issue for the accepted duplicate pair because it was filed earlier, has more context, and already has inbound reference activity.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45310 is the best issue representative overall among the accepted duplicate edge because it names the concrete regression clearly and appears to be the original report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Both involve label handling, but one is `gather_for_metrics` dropping tuple labels and the other is `reduce_labels` in `BeitImageProcessorFast`; different code paths and bugs." + }, + { + "left": "issue:45588", + "right": "issue:45859", + "accept": false, + "reason": "Both touch flash-attention/CPU-offload areas, but one is an `AttributeError` in `integrations/flash_attention.py` for sink-less models and the other is a rotary embedding incompatibility in Qwen3.5 MoE." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Both are import-related, but they concern different missing symbols (`_flash_supports_window_size` vs `HybridCache`) and different failure modes." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "These are unrelated runtime crashes: one is a device mismatch with CPU offload for `suno/bark-small`, the other is an AMP/CUDA loss crash in `LwDetrImageLoss`." + }, + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "Both mention mismatches, but one is a `LasrFeatureExtractor` center-argument regression and the other is a `device_map=\"auto\"` forward device mismatch in Gemma4 PLE." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "One is a SigLIP2 docs/example problem, the other is a broad position-id key warning across many models; not the same concrete bug." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "A Pegasus output-difference regression and a Qwen3 modular recurrent-state reset bug are different model behaviors and code paths." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "Both are model/export issues, but one is a German DistilBERT issue and the other is a TorchScript `GeneratorExp` export problem for ViT." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "The shared theme is image-processing label handling, but the SigLIP2 example errors and BeitImageProcessorFast `reduce_labels` bug are not the same defect." + }, + { + "left": "issue:45290", + "right": "issue:46326", + "accept": false, + "reason": "Both concern chat templating, but one crashes on tool-call messages with no content while the other is about `continue_final_message` for mistral_common tokenizers." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "These are separate model-loading/embedding issues: Llama 3.2 Vision load failure versus token-embedding resize reinitialization when `tie_word_embedding=False`." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both are `apply_chat_template` bugs, but one crashes on batched input with `padding=False` and the other returns zero assistant masks for multimodal inputs." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/loading regressions, but one is about unpacking BitNet weights during accelerate loading and the other is a stale GraniteSpeech device override test." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "A `resume_from_checkpoint` key mismatch is unrelated to the Qwen3 Omni MoE feature-extraction length helper behaving unexpectedly." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Both involve numeric precision issues, but the failing components and scenarios differ: `LwDetrImageLoss` under float16 AMP versus SwitchTransformers/TimmWrapperModel bfloat16 inference mismatches." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "Different problems: one is a LayoutLMv3 box-validation error message, the other is a Trainer `_gpu_broadcast_one` NoneType issue." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Different regressions in different subsystems: Pegasus output changes versus DeepSeek tokenizer behavior changes." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Both are runtime failures, but the causes differ completely: Trainer broadcast of `NoneType` versus MacOS bus error with a CLIP community model." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "These are distinct compile/runtime bugs: Bamba SDPA with `torch.compile` versus Qwen3 Omni MoE feature-length calculation." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 and unexpected position-id keys are different symptoms with no clear shared code-path." + }, + { + "left": "issue:42490", + "right": "issue:43575", + "accept": false, + "reason": "A local-path trailing-slash import warning and a tensor-parallel OOM when loading Qwen2-57B-A14B-Instruct are unrelated." + }, + { + "left": "issue:37428", + "right": "issue:42175", + "accept": false, + "reason": "One is a missing internal flash-attention helper import, the other is packaging/back-end dependency coverage for TensorFlow under `pip install .[torch]`." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "An embedding dtype drift during evaluate and a pipeline error message incorrectly implying translation support are different issues." + }, + { + "left": "issue:27957", + "right": "issue:34689", + "accept": false, + "reason": "Both involve model support changes, but XLMRoberta Flash Attention 2 and Llama 3.2 Vision loading failures are different problems." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "A dtype-change regression and a multimodal token-count helper using undefined video variables are unrelated bugs." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "These reports have the same title, same model family, and the same concrete `from_pretrained` regression for transformers>=5.4.0." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline `top_k` nesting behavior and Trainer GPU broadcast of `NoneType` are different features and failure modes." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Failing Flash Attention 2 tests and token-embedding reinitialization on resize are unrelated; one is test/FA2 related and the other is model initialization logic." + }, + { + "left": "issue:45593", + "right": "issue:47000", + "accept": false, + "reason": "Both are detector training losses, but one is D-FINE losing auxiliary losses when denoising is off and the other is RTDetrLoss infeasible cost matrix." + }, + { + "left": "issue:45459", + "right": "issue:46234", + "accept": false, + "reason": "Tokenizer protobuf error handling and a missing `is_falcon_mamba_ssm_available` utility export are unrelated import/exception issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is conservative and grounded in the packet. The only accepted edge, issue:45310 \u2194 issue:45478, is strongly supported by identical titles and matching Qwen3.5 MoE from_pretrained regression wording. The rejected pairs are generally framed cautiously as related symptoms rather than duplicates, which fits the evidence available." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 35, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:26413", + "issue:29870", + "issue:29925", + "issue:29942", + "issue:30064", + "issue:31501", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34689", + "issue:39692", + "issue:40444", + "issue:42490", + "issue:43257", + "issue:43278", + "issue:43299", + "issue:43381", + "issue:43526", + "issue:43701", + "issue:43873", + "issue:43927", + "issue:44186", + "issue:44360", + "issue:44448", + "issue:44464", + "issue:44466", + "issue:44530", + "issue:44561", + "issue:44589", + "issue:44779", + "issue:44855", + "issue:44977", + "issue:45127", + "issue:45198", + "issue:45229", + "issue:45245", + "issue:45290", + "issue:45381", + "issue:45588", + "issue:45698", + "issue:45823", + "issue:45920", + "issue:46192", + "issue:46399", + "issue:46424", + "issue:46655", + "issue:47276", + "issue:47436" + ], + "soft_pairs": [ + "issue:26413|issue:31501", + "issue:29870|issue:32090", + "issue:45127|issue:45245", + "issue:44360|issue:44561", + "issue:45290|issue:45381", + "issue:32090|issue:33290", + "issue:44448|issue:44977", + "issue:40444|issue:43526", + "issue:29870|issue:33290", + "issue:44589|issue:47436", + "issue:44464|issue:46424", + "issue:44530|issue:47276", + "issue:43278|issue:43381", + "issue:44855|issue:46399", + "issue:29925|issue:34689", + "issue:42490|issue:43873", + "issue:43927|issue:45698", + "issue:44977|issue:45229", + "issue:43257|issue:43299", + "issue:45920|issue:46192", + "issue:45290|issue:46655", + "issue:29942|issue:30064", + "issue:33357|issue:39692", + "issue:44561|issue:44779", + "issue:44186|issue:45198", + "issue:45588|issue:45823", + "issue:43873|issue:45229", + "issue:44448|issue:44466", + "issue:43381|issue:43701" + ], + "prepared_review_unit_hash": "08aec35d81c8e66774411e377124646c74de57012bbe67d430ec899069cfdbb7", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36438, + "estimated_input_tokens": 9110, + "estimated_eval_tokens": 18476 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the issues span unrelated training, generation, tokenizer, model-loading, quantization, and documentation problems. None of the soft edge candidates look like the same underlying bug or change, so no duplicates are accepted. If a single issue must represent the cluster, the broadest and most central is the v5 tokenizer regression in issue 45920.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 45920 is the broadest, most cross-cutting problem in the set: it describes a v5 regression affecting AutoTokenizer across multiple model families, which makes it the best representative of the cluster\u2019s dominant compatibility/regression theme.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45920 is the most suitable global issue because it is the most general and impactful regression among the listed issues, with multiple affected model families and clear evidence of broader fallout.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26413", + "right": "issue:31501", + "accept": false, + "reason": "Both are training-start failures, but the specific symptoms differ: resume-from-checkpoint epoch_iterator failure versus FineWeb SLM startup not beginning. No shared code-path is evident." + }, + { + "left": "issue:29870", + "right": "issue:32090", + "accept": false, + "reason": "Different failure modes in different parts of the stack: BART generate decoder_attention_mask assignment bug versus Trainer GPU broadcast NoneType error." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "One is a LoRA merge / tied-weights collapse issue, the other is a category-cardinality runtime limit. They are unrelated." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "A DSA indexer/ReLU discussion and a v5 utility-removal compatibility break are not the same bug or fix." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat template tool-call handling and Qwen2.5-VL video position IDs are distinct multimodal bugs with different code paths." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Unsupported NoneType GPU broadcast and Adafactor/DeepSpeed OOM are separate Trainer/optimizer issues." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Both are regression reports, but they target different models and different subsystems: Pegasus output drift versus Qwen3.5 flash-attention generation." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Multiple-images-per-prompt IterableDataset failure and BeitImageProcessorFast label reduction are unrelated image-processing bugs." + }, + { + "left": "issue:29870", + "right": "issue:33290", + "accept": false, + "reason": "A generate-time UnboundLocalError and a DeepSpeed OOM are clearly different problems." + }, + { + "left": "issue:44589", + "right": "issue:47436", + "accept": false, + "reason": "Float8 storage lookup failure and missing max_position_embeddings on Qwen2_5OmniConfig are unrelated runtime/config errors." + }, + { + "left": "issue:44464", + "right": "issue:46424", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and cache max length being honored only for quantized cache are separate cache/generation behaviors." + }, + { + "left": "issue:44530", + "right": "issue:47276", + "accept": false, + "reason": "A PagedAttentionCache crash and a misleading warning message are not the same underlying bug." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Embedding dtype drift during evaluation and gradient checkpointing being disallowed in eval mode are different eval-time issues." + }, + { + "left": "issue:44855", + "right": "issue:46399", + "accept": false, + "reason": "Both mention decorator-related failures, but they involve different models and different causes: Python 3.13 parsing versus missing decorator registration for rotary embeddings." + }, + { + "left": "issue:29925", + "right": "issue:34689", + "accept": false, + "reason": "SigLIP tokenizer fast-mode enforcement and Llama 3.2 Vision model-loading regression are not the same bug." + }, + { + "left": "issue:42490", + "right": "issue:43873", + "accept": false, + "reason": "Trailing-slash import warning and quantization offloading problems are unrelated." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/load behavior, but one is config token IDs being lost while the other is loading the wrong custom module. Different objects and failure modes." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3.5 flash-attention generation issues and Gemma4 multi-GPU CUDA OOM are separate model/runtime problems." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both concern Qwen3 MOE loading, but one is accelerate+deepspeed weight conversion and the other is a v5 loading break for Qwen3VL MOE models. Too different to merge." + }, + { + "left": "issue:45920", + "right": "issue:46192", + "accept": false, + "reason": "Wrong token IDs across multiple models is a tokenizer regression; silently ignored return_tensors with text_kwargs is a separate API-argument bug." + }, + { + "left": "issue:45290", + "right": "issue:46655", + "accept": false, + "reason": "Chat template tool-call crashes and SmolVLM video padding calculation are unrelated multimodal bugs." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures and void segmentation-map processing are unrelated test/image-processing issues." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "A MacOS bus error with a CLIP model and documentation example mistakes for SigLIP2 are not the same underlying issue." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 regressions, but one is a missing utility breaking trust_remote_code and the other is a tokenizer correctness regression. Different fixes." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2 tokenizer crashes and Wav2Vec2 save_pretrained/tokenization failures involve different model families and code paths." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "A flash_attention.py AttributeError on sink-less models and a Gemma4 device_map mismatch during forward are distinct runtime issues." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization offloading not working and multi-GPU CUDA OOM are related only at a very broad memory level, not the same bug." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Pegasus output differences between versions and lm_head.weight serialization differences depending on device are separate v5 regression reports." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are different checkpointing/training issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 36, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:23647", + "issue:29127", + "issue:29870", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:41720", + "issue:42490", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43278", + "issue:43334", + "issue:43381", + "issue:43412", + "issue:43531", + "issue:43575", + "issue:43611", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43976", + "issue:43994", + "issue:44077", + "issue:44206", + "issue:44265", + "issue:44361", + "issue:44387", + "issue:44442", + "issue:44479", + "issue:44568", + "issue:44743", + "issue:44871", + "issue:44977", + "issue:45072", + "issue:45198", + "issue:45229", + "issue:45372", + "issue:45412", + "issue:45440", + "issue:45491", + "issue:45563", + "issue:45588", + "issue:45803", + "issue:46121", + "issue:46421", + "issue:46858", + "issue:47276" + ], + "soft_pairs": [ + "issue:23647|issue:29870", + "issue:44871|issue:44977", + "issue:44361|issue:45198", + "issue:43278|issue:43575", + "issue:29127|issue:34567", + "issue:46421|issue:46858", + "issue:44206|issue:44479", + "issue:41720|issue:45229", + "issue:43412|issue:45588", + "issue:44265|issue:45198", + "issue:43653|issue:44442", + "issue:43611|issue:46121", + "issue:29870|issue:34689", + "issue:43976|issue:45440", + "issue:45198|issue:45803", + "issue:44077|issue:47276", + "issue:44977|issue:45563", + "issue:43334|issue:43531", + "issue:43653|issue:44743", + "issue:42915|issue:42994", + "issue:43065|issue:43994", + "issue:43575|issue:43701", + "issue:43756|issue:45440", + "issue:43575|issue:44387", + "issue:44568|issue:45491", + "issue:42490|issue:43381", + "issue:29127|issue:33357", + "issue:43994|issue:45412", + "issue:43065|issue:45412", + "issue:44442|issue:45072", + "issue:45372|issue:45588" + ], + "prepared_review_unit_hash": "4ad3e3d572b8d1a7f6b79f7f75523ce7f165bf68a782144fb32140e3443cc157", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37124, + "estimated_input_tokens": 9281, + "estimated_eval_tokens": 18818 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab-bag of unrelated Transformers issues across tokenizers, loading, generation, export, cache behavior, and model-specific regressions. The soft-similarity links are mostly superficial, so none of the candidate pairs should be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue: the items span unrelated bug classes and code paths, so there is no true duplicate cluster anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44442 is the most self-contained and reproducible standalone bug report in the set, so it is the best representative if one issue must be picked, but it does not unify the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:23647", + "right": "issue:29870", + "accept": false, + "reason": "Unrelated bugs: W&B sweeps custom objective integration vs BART generate() decoder_attention_mask initialization error." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Different model paths and failures: Gemma eos_token_id config mismatch vs Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "Different tokenizers/models and symptoms: MLuke task AttributeError vs Wav2Vec2 save_pretrained/tokenization failure." + }, + { + "left": "issue:43278", + "right": "issue:43575", + "accept": false, + "reason": "Both mention loading/training behavior, but one is dtype drift in eval and the other is tensor-parallel OOM on model load." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error messaging vs TrainerState token-count tracking are unrelated code paths." + }, + { + "left": "issue:46421", + "right": "issue:46858", + "accept": false, + "reason": "Both involve cache/compile behavior, but one is lazy cache init causing recompilation and the other is a device-tensor seq-length arithmetic bug." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Different regressions: unsupported feature-extractor arg vs video-input regression across Qwen variants." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Both are OOM/assertion-type failures, but they affect different models and likely different memory/device paths." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "Export-time Gemma3n issues vs flash-attention runtime crash with s_aux=None are distinct failures." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failure." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "Tokenizer special-token registration bug vs AutoTokenizer loading failure for a different tokenizer/model." + }, + { + "left": "issue:43611", + "right": "issue:46121", + "accept": false, + "reason": "Model loading with base_model_prefix vs rope-validation helper TypeError are separate mechanisms." + }, + { + "left": "issue:29870", + "right": "issue:34689", + "accept": false, + "reason": "BART generation decoder mask bug vs Llama 3.2 Vision loading break are different model/version regressions." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility issue vs DeepSeekV3 implementation divergence are unrelated." + }, + { + "left": "issue:45198", + "right": "issue:45803", + "accept": false, + "reason": "Wav2Vec2 tokenization/save failure vs bare-except swallowing interrupts in FuyuBatchFeature are different bugs." + }, + { + "left": "issue:44077", + "right": "issue:47276", + "accept": false, + "reason": "PatchTSMixer post_init policy change vs synced_gpus warning wording are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Generation failure on Qwen3.5 vs stale warning in paged generate() are different issues." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Qwen3-VL missing pad_token_id and Qwen3-MoE sliding_window behavior are separate model-specific bugs." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBird tokenizer special-token registration vs Qwen3 recurrent-state reset are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Qwen3Moe FP8 config failure vs quantized model saving failure are different code paths." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder vs Siglip2 nonsensical outputs are unrelated model problems." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Tensor-parallel OOM on Qwen2 load vs resume_from_checkpoint key mismatch are different failure modes." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Smollm3 RoPE-layer discrepancy vs DeepSeekV3 divergence are unrelated architecture/config issues." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both are memory complaints, but one is TP load OOM and the other is increased reserved memory under int4 quantization." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer add_special_tokens regression vs Gemma3 sliding-window NaN embeddings are different bugs." + }, + { + "left": "issue:42490", + "right": "issue:43381", + "accept": false, + "reason": "Local-path import warning and gradient-checkpointing-in-eval are unrelated." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 error clarity vs CLIP bus error on macOS are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "Siglip2 loading/output issue vs RT-DETR memory leak are different model classes and symptoms." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "Sam3PixelDecoder dummy layer vs RT-DETR deletion/GC memory leak are unrelated." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "AutoTokenizer loading failure vs dtype-mismatch inference bug in SwitchTransformers/TimmWrapperModel are different code paths." + }, + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Gemma 4 processor import failure from mistral_common vs flash_attention.py crash on sink-less models are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 37, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:23647", + "issue:27957", + "issue:28282", + "issue:29127", + "issue:29870", + "issue:29925", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:42915", + "issue:43066", + "issue:43381", + "issue:43421", + "issue:43526", + "issue:43824", + "issue:43873", + "issue:43931", + "issue:43994", + "issue:44186", + "issue:44279", + "issue:44442", + "issue:44462", + "issue:44514", + "issue:44625", + "issue:44857", + "issue:44898", + "issue:45245", + "issue:45250", + "issue:45372", + "issue:45381", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45712", + "issue:45834", + "issue:46369", + "issue:46640", + "issue:46693", + "issue:46701", + "issue:46821", + "issue:47217" + ], + "soft_pairs": [ + "issue:45491|issue:46693", + "issue:43526|issue:47217", + "issue:45406|issue:46701", + "issue:44625|issue:45584", + "issue:29127|issue:35141", + "issue:45440|issue:45593", + "issue:45459|issue:45712", + "issue:43421|issue:45491", + "issue:43526|issue:45834", + "issue:45250|issue:46640", + "issue:45372|issue:45406", + "issue:45412|issue:46369", + "issue:43824|issue:44279", + "issue:33290|issue:34689", + "issue:33290|issue:35141", + "issue:43381|issue:43873", + "issue:27957|issue:33290", + "issue:45245|issue:45563", + "issue:45372|issue:45459", + "issue:43526|issue:44186", + "issue:43994|issue:46369", + "issue:44857|issue:45588", + "issue:23647|issue:28282", + "issue:42915|issue:43931", + "issue:43066|issue:44462", + "issue:29870|issue:29925", + "issue:45588|issue:46821", + "issue:27957|issue:29127", + "issue:44442|issue:44898", + "issue:44514|issue:45381" + ], + "prepared_review_unit_hash": "d2c47051c532ce9590f5cedbddc6bd026b41862578d685b71d70723bdc38dab1", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35929, + "estimated_input_tokens": 8983, + "estimated_eval_tokens": 18222 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose mix of unrelated Transformers issues spanning tokenizers, vision processors, generation, Flash Attention, quantization, and model-loading. None of the soft pairs look like the same underlying bug, so they should stay separate.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical duplicate is evident; if one issue must anchor the cluster, issue 43873 is the broadest/highest-activity open item, but it does not actually subsume the others.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43873 is the best cluster anchor because it is active, open, and broad ('offloading' + quantization), but it is still not a good representative duplicate for the mixed set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45491", + "right": "issue:46693", + "accept": false, + "reason": "Different failures: NaN embeddings in mixed-length batching vs FA2 long-context performance regression." + }, + { + "left": "issue:43526", + "right": "issue:47217", + "accept": false, + "reason": "Different subsystems: BeitImageProcessorFast reduce_labels vs VL image-placeholder handling in user text." + }, + { + "left": "issue:45406", + "right": "issue:46701", + "accept": false, + "reason": "Different Gemma-related bugs: missing _tokenizer in serve vs image-text pipeline output issue." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Different models and code paths: Qwen3.5 num_labels propagation vs Whisper empty-transcription failure after align_special_tokens." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "Different concerns: LayoutLMv3 error-message clarity vs embedding resize reinitialization bug." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "Different models/features: DeepseekV3MoE remote divergence vs D-FINE auxiliary-loss omission." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Different issues: protobuf exception masking tokenizer errors vs leftover dummy classes in dummy_pt_objects.py." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Different bugs: tokenizer post-processor not updating runtime special tokens vs Gemma3 NaN embeddings." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "Different issues: Beit reduce_labels returning one label vs Kosmos2.5 long OCR input index error." + }, + { + "left": "issue:45250", + "right": "issue:46640", + "accept": false, + "reason": "Too vague, but these are still separate FA2 reports with different scope and symptoms." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Different bugs: ReasoningEffort import error breaking Gemma4 loading vs Gemma4Processor _tokenizer attribute error in serve." + }, + { + "left": "issue:45412", + "right": "issue:46369", + "accept": false, + "reason": "Different memory problems: RT-DETR GC leak vs SAM3-litetext memory overflow." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "Different issues: Qwen2.5-VL import failure vs generic dependency issue with transformers." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Different bugs: deepspeed Adafactor OOM vs Llama 3.2 Vision model-loading regression." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "Different bugs: optimizer OOM vs embedding resize reinitialization." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Different behaviors: eval-mode gradient-checkpointing prohibition vs quantization offloading failure." + }, + { + "left": "issue:27957", + "right": "issue:33290", + "accept": false, + "reason": "Different models and failures: XLMRoberta Flash Attention 2 vs deepspeed Adafactor OOM." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Different issues: category-count limit vs stale generate warning for num_return_sequences." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Different bugs: import error for ReasoningEffort vs protobuf-specific tokenizer error masking." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Different tokenizer/image-processor bug vs LayoutLMv2Tokenizer NER padding/truncation crash." + }, + { + "left": "issue:43994", + "right": "issue:46369", + "accept": false, + "reason": "Different models and symptoms: SigLIP2 wrong outputs vs SAM3 memory overflow." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "Different code paths: RT-DETR image loss AMP crash vs flash_attention sink-less AttributeError." + }, + { + "left": "issue:23647", + "right": "issue:28282", + "accept": false, + "reason": "Different areas: wandb sweeps custom objective vs missing PyTorch ImportError." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "Different Qwen model failures: FineGrainedFP8Config failure vs Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Related tokenizer area, but different concrete bugs: wrong decoder type in v5 vs ignoring tokenizer.json." + }, + { + "left": "issue:29870", + "right": "issue:29925", + "accept": false, + "reason": "Different issues: BART generate decoder-mask bug vs SigLIP fast-tokenizer enforcement." + }, + { + "left": "issue:45588", + "right": "issue:46821", + "accept": false, + "reason": "Different subsystems: flash_attention sink-less crash vs TimesFM window_size AttributeError." + }, + { + "left": "issue:27957", + "right": "issue:29127", + "accept": false, + "reason": "Different models and code paths: XLMRoberta Flash Attention 2 vs LayoutLMv3 error-message clarity." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "Different failures: tokenizer loading vs Perceiver non-default-resolution failure." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Same model family, but different concrete bugs: batched apply_chat_template crash vs wrong video vision_position_ids." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 38, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:28282", + "issue:29870", + "issue:42175", + "issue:42490", + "issue:42617", + "issue:43262", + "issue:43316", + "issue:43493", + "issue:43525", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43643", + "issue:43646", + "issue:43701", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:44077", + "issue:44206", + "issue:44442", + "issue:44479", + "issue:44743", + "issue:44811", + "issue:44855", + "issue:44871", + "issue:44913", + "issue:44964", + "issue:44977", + "issue:44987", + "issue:45137", + "issue:45245", + "issue:45440", + "issue:45759", + "issue:45859", + "issue:46121", + "issue:46178", + "issue:46234", + "issue:46399", + "issue:46821", + "issue:47312", + "issue:47436" + ], + "soft_pairs": [ + "issue:43646|issue:44077", + "issue:46821|issue:47436", + "issue:43316|issue:47312", + "issue:46121|issue:46399", + "issue:43262|issue:44206", + "issue:43761|issue:45859", + "issue:43720|issue:44442", + "issue:25251|issue:28282", + "issue:44743|issue:44811", + "issue:42175|issue:46234", + "issue:44077|issue:44479", + "issue:44977|issue:45245", + "issue:43873|issue:44871", + "issue:42490|issue:44743", + "issue:43316|issue:45759", + "issue:43493|issue:43827", + "issue:42617|issue:43575", + "issue:43749|issue:43931", + "issue:43643|issue:44913", + "issue:44987|issue:46178", + "issue:43550|issue:43701", + "issue:27957|issue:29870", + "issue:43316|issue:43525", + "issue:43606|issue:43701", + "issue:44964|issue:45440", + "issue:43575|issue:43742", + "issue:44855|issue:45137" + ], + "prepared_review_unit_hash": "dcb5c63d8c10f3e3720cb68611e8325534d7109a2555acb9f873d08c4e1519df", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35789, + "estimated_input_tokens": 8948, + "estimated_eval_tokens": 18152 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous regression/bug reports with only superficial similarity by subsystem or wording. None of the soft pairs look like the same underlying bug/change, so no duplicate merges are warranted.", + "confidence": 0.94, + "canonical_issue_reason": "No clear canonical issue: the soft pairs are mostly unrelated, and the closest match (issue:43646) is only a broad regression report rather than a true duplicate hub.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue:43646 is the closest thing to an umbrella report because it is broad and regression-like, but the cluster is too mixed to collapse into one duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Both touch custom model init, but one is a broad v5 regression while the other is a patchtsmixer-specific post_init rule; not the same bug." + }, + { + "left": "issue:46821", + "right": "issue:47436", + "accept": false, + "reason": "Different models and different missing attributes (`window_size` vs `max_position_embeddings`); same error class, not same issue." + }, + { + "left": "issue:43316", + "right": "issue:47312", + "accept": false, + "reason": "Gemma3TextConfig API mismatch and kernels 0.16 incompatibility are unrelated problems." + }, + { + "left": "issue:46121", + "right": "issue:46399", + "accept": false, + "reason": "Rope-parameter JSON/list handling is a different code path from the Qwen3.5 kernel registration crash." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate default and unsupported `center` arg in a feature extractor are distinct audio bugs." + }, + { + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "CLIPVision hidden-state regression and Qwen3.5 MoE CPU-offload incompatibility are different failures." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "BitNet packed-weight loading and FastSpeech2Conformer tokenizer loading are unrelated model-loading problems." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline `top_k` output shape bug and missing PyTorch ImportError are completely different." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Qwen3.5 recurrent-state reset and Whisper `batch_decode` special-token handling are separate behaviors." + }, + { + "left": "issue:42175", + "right": "issue:46234", + "accept": false, + "reason": "Packaging/backend omission and a missing utility symbol are not the same bug." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "PatchTSMixer `post_init` validation and Qwen video-input regression do not share a concrete code path." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Flash-attention generation issues and category-count overflow are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantized offloading behavior and Gemma eos-token config inconsistency are different issues." + }, + { + "left": "issue:42490", + "right": "issue:44743", + "accept": false, + "reason": "Local-path import warnings and Qwen recurrent-state handling are unrelated." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Both are config-related, but one is a Gemma3TextConfig API discrepancy and the other is a multimodal `text_config` unwrapping bug." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation drift and outdated docs referencing `pipeline()` are not the same underlying change." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "3d_parallel runtime failure and tensor-parallel OOM are different loading/execution problems." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU-RAM-efficient loading breakage and Qwen3-VL shape mismatch are separate model-loading bugs." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "Remote-code config field loss and GPTNeoX rotary_pct not persisting on reload are different config persistence issues." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "Transformers>=5.1 loading failure and tensorflow_text backend registration are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba and resume checkpoint key mismatch do not share a code path." + }, + { + "left": "issue:27957", + "right": "issue:29870", + "accept": false, + "reason": "Flash Attention 2 support for XLMRoberta and BART generate UnboundLocalError are unrelated." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Both involve config APIs, but the missing/variant fields are model-specific and not the same bug." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch and checkpoint-key mismatch are different loading errors." + }, + { + "left": "issue:44964", + "right": "issue:45440", + "accept": false, + "reason": "Phi-4 multimodal loading failure and DeepSeekV3MoE divergence from remote implementation are not the same problem." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "Tensor-parallel OOM and MobileLLM key error are distinct model-loading issues." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 import parsing failure and DeepSpeed ZeRO3 deque underflow are unrelated runtime bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 39, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:23647", + "issue:25251", + "issue:27957", + "issue:28218", + "issue:32090", + "issue:43065", + "issue:43232", + "issue:43295", + "issue:43316", + "issue:43388", + "issue:43452", + "issue:43646", + "issue:43701", + "issue:43716", + "issue:43761", + "issue:43825", + "issue:43931", + "issue:43950", + "issue:44062", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44190", + "issue:44442", + "issue:44464", + "issue:44496", + "issue:44521", + "issue:44534", + "issue:44589", + "issue:44617", + "issue:44871", + "issue:44913", + "issue:45216", + "issue:45584", + "issue:45803", + "issue:45907", + "issue:46050", + "issue:46234", + "issue:46291", + "issue:46421", + "issue:46506", + "issue:46829", + "issue:46932", + "issue:47217", + "issue:47276", + "issue:47427" + ], + "soft_pairs": [ + "issue:43950|issue:47427", + "issue:44062|issue:45907", + "issue:43761|issue:44079", + "issue:43701|issue:46506", + "issue:44190|issue:44496", + "issue:43316|issue:43716", + "issue:43931|issue:45216", + "issue:46829|issue:46932", + "issue:27957|issue:32090", + "issue:44442|issue:45803", + "issue:44589|issue:47276", + "issue:43295|issue:44871", + "issue:23647|issue:28218", + "issue:44871|issue:45584", + "issue:43295|issue:43825", + "issue:16998|issue:22355", + "issue:43761|issue:44188", + "issue:46234|issue:46291", + "issue:43232|issue:43388", + "issue:43452|issue:44496", + "issue:44521|issue:47217", + "issue:44077|issue:44534", + "issue:43646|issue:44913", + "issue:44871|issue:46050", + "issue:25251|issue:27957", + "issue:43065|issue:44617", + "issue:44464|issue:46421" + ], + "prepared_review_unit_hash": "77e41233528be6393571f89e132bd232764a6ae408e2a7cb472a3e1d8e785a5f", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35741, + "estimated_input_tokens": 8936, + "estimated_eval_tokens": 18128 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a grab-bag of unrelated issue reports connected only by broad surface similarity (generation, tokenizers, multimodal, config/loading, etc.). None of the soft pairs look like the same underlying bug/change, so they should not be merged as duplicates.", + "confidence": 0.92, + "canonical_issue_reason": "If a single anchor is needed, #44521 is the strongest representative because it is open, has the most inbound references, and describes a concrete reproducible multimodal regression; however, the overall cluster is too heterogeneous to treat as one duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "#44521 is the best standalone issue in this set: it is active, well-scoped, and has the clearest user-facing reproduction. It does not, however, unify the rest of the issues into one bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43950", + "right": "issue:47427", + "accept": false, + "reason": "Both touch initialization behavior, but one is about corrupting non-persistent buffers during load while the other is about skipping redundant parameter init on non-CUDA hardware; different bugs and code paths." + }, + { + "left": "issue:44062", + "right": "issue:45907", + "accept": false, + "reason": "`AddedToken()` argument handling and `list_repo_files` revision forwarding are unrelated API issues in different subsystems." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "One is a CLIPVisionModel regression returning `hidden_states=None`; the other is a `ModelOutput` key assignment bug. Similar surface area, different failure." + }, + { + "left": "issue:43701", + "right": "issue:46506", + "accept": false, + "reason": "Checkpoint key mismatches in resume logic and CPU/CUDA output differences in PPDocLayoutV3 are unrelated." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Loading a local dataset in a training script and failing to recognize a model config are different loading problems with different causes." + }, + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "Gemma3TextConfig API inconsistency and Mistral-3 image preprocessor dtype mismatch are different multimodal issues." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Both involve Qwen/VL model loading/saving, but one is a weight-shape mismatch on load and the other is an incorrect saved checkpoint in `save_pretrained`; not the same concrete bug." + }, + { + "left": "issue:46829", + "right": "issue:46932", + "accept": false, + "reason": "VideoPrism hidden-state field corruption and Gemma4 quantized embedder dtype casting are unrelated model-output vs quantization bugs." + }, + { + "left": "issue:27957", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 support for XLMRoberta and a Trainer GPU broadcast `NoneType` error are unrelated." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "FastSpeech2Conformer tokenizer loading and Fuyu batch tensor conversion swallowing exceptions are different tokenizer/data-path issues." + }, + { + "left": "issue:44589", + "right": "issue:47276", + "accept": false, + "reason": "Missing Float8 storage handling and a misleading synced_gpus warning are not the same bug; one is a hard runtime failure, the other is a message issue." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "A regression in `processor.tokenizer` / image passthrough is unrelated to inconsistent `eos_token_id` configuration." + }, + { + "left": "issue:23647", + "right": "issue:28218", + "accept": false, + "reason": "WandB sweep custom objective support and tokenizer-added spacing after added tokens have no shared code path." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Both touch generation/tokenizer behavior, but one is an EOS-token config mismatch and the other is empty-transcription failure after special-token alignment; distinct defects." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "A custom model API regression and an error-message wording bug in `pipeline()` are not duplicates." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "`model_max_length` behavior in DeBERTa-V3 and a missing `transformers.onnx` module are unrelated." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "Hidden states being dropped in CLIPVisionModel is unrelated to attention-kernel divergence under `torch.compile`." + }, + { + "left": "issue:46234", + "right": "issue:46291", + "accept": false, + "reason": "A missing utility flag in `import_utils` and a kernels import failure due to version/revision requirements are different import-time issues." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs updates after `sync_gpus` and `gather_for_metrics` truncating tuple labels are different distributed-training bugs." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "`gguf_file` breaking `from_pretrained` for tokenizer/model loading is unrelated to an unrecognized model config in `AutoModel`." + }, + { + "left": "issue:44521", + "right": "issue:47217", + "accept": false, + "reason": "Both involve multimodal chat/template handling, but one is about assistant masks being all zero and the other is about image placeholders in user text causing model failure; not clearly the same code-path bug." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "Optional `post_init` deprecation/validation and non-persistent buffer corruption are separate initialization problems." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "A broad Transformers 5.0 custom model-init breakage and GPTNeoXConfig `rotary_pct` not persisting on reload are different configuration/init bugs." + }, + { + "left": "issue:44871", + "right": "issue:46050", + "accept": false, + "reason": "`eos_token_id` inconsistency in Gemma-3 and sinq quantized model save/load breakage are different save/load regressions." + }, + { + "left": "issue:25251", + "right": "issue:27957", + "accept": false, + "reason": "Pipeline top-k nesting behavior and XLMRoberta Flash Attention 2 support are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "A dummy `nn.Conv2d` in Sam3PixelDecoder and Sam3Video CUDA OOM are different SAM3 issues with different failure modes." + }, + { + "left": "issue:44464", + "right": "issue:46421", + "accept": false, + "reason": "Chunked generation inconsistencies under compiled forward and lazy cache init causing recompilation are related by theme but not the same concrete bug or fixable in one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 40, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:23647", + "issue:27957", + "issue:29925", + "issue:35141", + "issue:41950", + "issue:42490", + "issue:42907", + "issue:43352", + "issue:43404", + "issue:43441", + "issue:43475", + "issue:43504", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43931", + "issue:44060", + "issue:44162", + "issue:44479", + "issue:44514", + "issue:44534", + "issue:44610", + "issue:44792", + "issue:44855", + "issue:44913", + "issue:44977", + "issue:45127", + "issue:45200", + "issue:45381", + "issue:45405", + "issue:45542", + "issue:45588", + "issue:45600", + "issue:45632", + "issue:45698", + "issue:45803", + "issue:46170", + "issue:46178", + "issue:46234", + "issue:46291", + "issue:46365", + "issue:46369", + "issue:46424", + "issue:46640", + "issue:46693", + "issue:47013", + "issue:47217" + ], + "soft_pairs": [ + "issue:43404|issue:43441", + "issue:42907|issue:45698", + "issue:43352|issue:46640", + "issue:44162|issue:44792", + "issue:46178|issue:46291", + "issue:43540|issue:44514", + "issue:44792|issue:44977", + "issue:44479|issue:47217", + "issue:45803|issue:46365", + "issue:44610|issue:46369", + "issue:45381|issue:47217", + "issue:44534|issue:44913", + "issue:44792|issue:45600", + "issue:45542|issue:46234", + "issue:27957|issue:29925", + "issue:43525|issue:43531", + "issue:43504|issue:45698", + "issue:35141|issue:42490", + "issue:23647|issue:27957", + "issue:46693|issue:47217", + "issue:45803|issue:46170", + "issue:43931|issue:44060", + "issue:45632|issue:46424", + "issue:44855|issue:45588", + "issue:45588|issue:47013", + "issue:45127|issue:45200", + "issue:44792|issue:45405", + "issue:41950|issue:43475" + ], + "prepared_review_unit_hash": "20805e9bcfd82a0bcec92e4b0fa7b97f2ff6853b5a2f4a218e7cbdee21c3fbe2", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35551, + "estimated_input_tokens": 8888, + "estimated_eval_tokens": 18032 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most items are distinct model- or subsystem-specific bugs, and the soft pairs are mostly superficial text similarity rather than the same underlying defect. I would not merge any of the proposed pairs as duplicates.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Same model family, but one is tied-weight handling and the other is FlashAttention support; different code paths and fixes." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/load flows, but one is about saving dequantized models and the other about loading the wrong custom module; not the same bug." + }, + { + "left": "issue:43352", + "right": "issue:46640", + "accept": false, + "reason": "Issue 43352 is a specific Nemotron FlashAttention2 support gap; 46640 is a broader FA2 issue with no evidence it is the same failure." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 breakage and a Janus image-generation test failure are unrelated bugs." + }, + { + "left": "issue:46178", + "right": "issue:46291", + "accept": false, + "reason": "Both are import/dependency failures, but they affect different packages and failure modes." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both are multimodal input issues, but they hit different processors/models and different shapes of failure." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Janus image generation and Qwen3.5 flash-attention generation problems are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:47217", + "accept": false, + "reason": "Both concern VL inputs, but one is a video-input regression and the other is image-placeholder handling in user text." + }, + { + "left": "issue:45803", + "right": "issue:46365", + "accept": false, + "reason": "Bare except swallowing errors in Fuyu is unrelated to the Trainer import issue caused by datasets." + }, + { + "left": "issue:44610", + "right": "issue:46369", + "accept": false, + "reason": "Processor image-size mismatch and SAM3 LiteText memory overflow are different bugs." + }, + { + "left": "issue:45381", + "right": "issue:47217", + "accept": false, + "reason": "Both are VL bugs, but one is about video position ids and the other is placeholder placement in chat text." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Serialization buffer junk and GPTNeoX rotary_pct reload drift are separate config/state issues." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "A Janus test failure is unrelated to stale auto_mappings references breaking CI." + }, + { + "left": "issue:45542", + "right": "issue:46234", + "accept": false, + "reason": "TensorBoard/TF backend handling and missing falcon-mamba availability helper are unrelated." + }, + { + "left": "issue:27957", + "right": "issue:29925", + "accept": false, + "reason": "XLMRoberta FlashAttention2 support and SigLIP use_fast enforcement are different model/processor issues." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config and Qwen3-MoE sliding_window behavior are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "Legacy-field preset loading and wrong custom module after save_pretrained are different loading bugs." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "Embedding reinitialization on resize and trailing-slash local path warnings do not share the same root cause." + }, + { + "left": "issue:23647", + "right": "issue:27957", + "accept": false, + "reason": "Wandb sweeps custom objective integration is unrelated to FlashAttention2 support." + }, + { + "left": "issue:46693", + "right": "issue:47217", + "accept": false, + "reason": "Long-context FA2 performance regression and VL placeholder parsing are different failures." + }, + { + "left": "issue:45803", + "right": "issue:46170", + "accept": false, + "reason": "One is exception handling in a converter; the other is a Windows/ROCm import dependency problem." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Both mention weights, but one is a load-time shape mismatch and the other is a tied-weights warning bug." + }, + { + "left": "issue:45632", + "right": "issue:46424", + "accept": false, + "reason": "Cache-path collisions for trust_remote_code and max_cache_length behavior are unrelated cache bugs." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "Python 3.13 decorator parsing and flash_attention s_aux=None crashes have no shared code path." + }, + { + "left": "issue:45588", + "right": "issue:47013", + "accept": false, + "reason": "Both crash at runtime, but in different layers and with different causes." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA merge collapse with tied embeddings and Gemma 4 mm_token_type_ids defaults are not the same defect." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "A Janus test failure is unrelated to an unreleased MIN_PEFT_VERSION bump." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Video-classification processor selection and SAM3 VisionEncoderOutput missing a field are different bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 41, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:23647", + "issue:25251", + "issue:29127", + "issue:29870", + "issue:35141", + "issue:39692", + "issue:41720", + "issue:43065", + "issue:43335", + "issue:43388", + "issue:43550", + "issue:43644", + "issue:43650", + "issue:43673", + "issue:43688", + "issue:43827", + "issue:43854", + "issue:43994", + "issue:44190", + "issue:44242", + "issue:44479", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44869", + "issue:44913", + "issue:44964", + "issue:45072", + "issue:45245", + "issue:45381", + "issue:45406", + "issue:45563", + "issue:45584", + "issue:46082", + "issue:46202", + "issue:46369", + "issue:46399", + "issue:46655", + "issue:46697", + "issue:46832", + "issue:46858", + "issue:46894", + "issue:46932", + "issue:47312", + "issue:47321" + ], + "soft_pairs": [ + "issue:43673|issue:44743", + "issue:44242|issue:46894", + "issue:46082|issue:46697", + "issue:43650|issue:46202", + "issue:44190|issue:44964", + "issue:44617|issue:45584", + "issue:25251|issue:29870", + "issue:43827|issue:43994", + "issue:43388|issue:43688", + "issue:29870|issue:35141", + "issue:45072|issue:46932", + "issue:45245|issue:46858", + "issue:43644|issue:44913", + "issue:44734|issue:45406", + "issue:44869|issue:46832", + "issue:43335|issue:44855", + "issue:46399|issue:47312", + "issue:23647|issue:29127", + "issue:43854|issue:44792", + "issue:45381|issue:46369", + "issue:39692|issue:47321", + "issue:43827|issue:47321", + "issue:41720|issue:45245", + "issue:43550|issue:44190", + "issue:45563|issue:45584", + "issue:44617|issue:46369", + "issue:43065|issue:44610", + "issue:44479|issue:44734", + "issue:45381|issue:46655" + ], + "prepared_review_unit_hash": "710280380da3cd1b73881f1232c60ee086c9f74ebc78e5bd62cf8574232e5532", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36514, + "estimated_input_tokens": 9129, + "estimated_eval_tokens": 18514 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft pairs share only broad topical similarity, not the same underlying bug or change. I found no true duplicate pair, so the cluster should not be merged on these edges.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43644 is the broadest, most cross-cutting regression in the set (framework-level buffer initialization behavior) and is more representative of the cluster than the many narrowly scoped model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the best single representative issue because it has the widest impact and clearest general runtime-regression framing among otherwise unrelated reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Different failures: a generic GenerationMixin cache/chunked_prefill regression vs Qwen3.5 recurrent states resetting with cache and seq_len>1." + }, + { + "left": "issue:44242", + "right": "issue:46894", + "accept": false, + "reason": "Different model/loss bugs: MoE router load-balancing loss omission vs CohereASR double-shift training-loss bug." + }, + { + "left": "issue:46082", + "right": "issue:46697", + "accept": false, + "reason": "Unrelated domains: LlamaConfig head_dim validation vs Swin2SR image padding in an image processor." + }, + { + "left": "issue:43650", + "right": "issue:46202", + "accept": false, + "reason": "No shared bug shape: a vague data request vs a Gemma pipeline roundtrip/tokenization issue." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Different load paths and symptoms: local dataset loading in a training script vs loading a multimodal model." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Different subsystems: Sam3Video OOM vs Whisper empty-transcription failure after align_special_tokens." + }, + { + "left": "issue:25251", + "right": "issue:29870", + "accept": false, + "reason": "Different pipeline/eval bugs: top_k nested-list output vs BART generate() unbound local error." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs referencing removed pipeline() APIs is not the same as SigLIP2 inference producing nonsensical results." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Different training/eval bugs: gather_for_metrics label truncation vs auxiliary-loss normalization." + }, + { + "left": "issue:29870", + "right": "issue:35141", + "accept": false, + "reason": "Different code paths: BART generate() assignment bug vs embedding resize/post_init weight reinitialization." + }, + { + "left": "issue:45072", + "right": "issue:46932", + "accept": false, + "reason": "Both mention dtype issues, but they affect different models and mechanisms; not the same concrete bug." + }, + { + "left": "issue:45245", + "right": "issue:46858", + "accept": false, + "reason": "Category-count limit error is unrelated to the static-cache slice arithmetic/device-tensor bug." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Different regressions: non-persistent buffer initialization junk vs GPTNeoXConfig reload dropping rotary_pct." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Both involve transformers serve, but one is KV-cache continuation indexing and the other is a missing _tokenizer attribute." + }, + { + "left": "issue:44869", + "right": "issue:46832", + "accept": false, + "reason": "Whisper word-timestamp decoding crash is unrelated to RT-DETR feature-level/backbone mismatch." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "Sparse-layer creation in SwitchTransformersConfig is unrelated to the Python 3.13 IndentationError/JIT parsing issue." + }, + { + "left": "issue:46399", + "right": "issue:47312", + "accept": false, + "reason": "Kernel compatibility concerns overlap only loosely; one is a missing decorator registration bug in Qwen3.5, the other is a general kernels version incompatibility report." + }, + { + "left": "issue:23647", + "right": "issue:29127", + "accept": false, + "reason": "Wandb sweep custom objective integration is unrelated to LayoutLMv3 box-information error handling." + }, + { + "left": "issue:43854", + "right": "issue:44792", + "accept": false, + "reason": "GLM-4.7-Flash loading problems are not the same as a janus image-generation test failure." + }, + { + "left": "issue:45381", + "right": "issue:46369", + "accept": false, + "reason": "Qwen2.5-VL video position IDs bug is unrelated to Sam3-litetext memory overflow." + }, + { + "left": "issue:39692", + "right": "issue:47321", + "accept": false, + "reason": "SigLIP2 doc/example issues and AMD quark class updates are unrelated." + }, + { + "left": "issue:43827", + "right": "issue:47321", + "accept": false, + "reason": "Docs still using pipeline() and AMD quark class updates are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "A800 cudaErrorAssert in Qwen3 auto device mapping is unrelated to the 2^24 category limit error." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba is unrelated to local dataset loading in a training script." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "Stale warning for num_return_sequences is unrelated to Whisper empty-transcription failure." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "Both are memory-related, but they are different models and different failure modes (OOM vs memory overflow)." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same bug as OmDet-Turbo processor/model input-size mismatch." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Video-input regression across Qwen models is unrelated to the serve KV-cache indexing crash." + }, + { + "left": "issue:45381", + "right": "issue:46655", + "accept": false, + "reason": "Both involve video processors, but one is Qwen2.5-VL vision_position_ids and the other is SmolVLM padding calculation." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 42, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:27957", + "issue:35141", + "issue:39692", + "issue:42175", + "issue:43295", + "issue:43317", + "issue:43335", + "issue:43526", + "issue:43540", + "issue:43611", + "issue:43643", + "issue:43673", + "issue:43854", + "issue:43994", + "issue:44112", + "issue:44190", + "issue:44242", + "issue:44246", + "issue:44279", + "issue:44462", + "issue:44496", + "issue:44521", + "issue:44734", + "issue:44779", + "issue:44933", + "issue:44964", + "issue:45003", + "issue:45127", + "issue:45276", + "issue:45357", + "issue:45362", + "issue:45375", + "issue:45381", + "issue:45520", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45907", + "issue:46170", + "issue:46421", + "issue:46489", + "issue:46617", + "issue:46697", + "issue:46858", + "issue:46897", + "issue:47145", + "issue:47312", + "issue:47321" + ], + "soft_pairs": [ + "issue:43526|issue:46697", + "issue:45632|issue:46617", + "issue:43295|issue:45276", + "issue:43335|issue:44190", + "issue:43643|issue:45375", + "issue:39692|issue:44190", + "issue:44933|issue:45362", + "issue:44779|issue:46489", + "issue:43854|issue:44964", + "issue:43540|issue:44734", + "issue:45276|issue:45584", + "issue:43994|issue:47321", + "issue:47145|issue:47321", + "issue:44242|issue:46897", + "issue:43643|issue:44496", + "issue:43673|issue:46421", + "issue:45127|issue:45357", + "issue:45520|issue:46170", + "issue:43611|issue:47321", + "issue:44521|issue:45381", + "issue:43317|issue:47312", + "issue:44112|issue:45600", + "issue:42175|issue:44279", + "issue:44112|issue:44964", + "issue:44246|issue:45003", + "issue:27957|issue:35141", + "issue:44462|issue:45907", + "issue:43673|issue:46858" + ], + "prepared_review_unit_hash": "fbf0671c2eb8d8919c39958331be706cdf89d5a7e39730354e91fd13b3b0c619", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36171, + "estimated_input_tokens": 9043, + "estimated_eval_tokens": 18342 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is largely heterogeneous: it mixes unrelated bugs in tokenizers, multimodal processors, cache handling, configs, and CI. Only two soft pairs look like the same underlying issue: the DeepSeek tokenizer regression and the trust_remote_code cache-loading failures.", + "confidence": 0.57, + "canonical_issue_reason": "No single coherent duplicate theme spans the whole set, so the best representative is the most active open issue, issue 44521, which has the strongest discussion/inbound signal and is broadly scoped around multimodal chat-template behavior.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44521 is the most suitable representative for the cluster overall because it is open, has the most engagement, and is the broadest live bug report among the items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43526", + "right": "issue:46697", + "accept": false, + "reason": "Different code paths: BeitImageProcessorFast label reduction vs Swin2SR image padding. No shared underlying bug." + }, + { + "left": "issue:45632", + "right": "issue:46617", + "accept": true, + "reason": "Both point to trust_remote_code cache/path resolution failures when loading models from cache; likely the same cache-path bug manifesting differently." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Different regression areas: processor/tokenizer access in custom multimodal code vs Gemma4 resize_token_embeddings not updating embeddings." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug and local dataset loading failure in an example script are unrelated." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "One is a trust_remote_code AutoConfig field-loss issue; the other is a specific config field being dropped by @strict. Similar theme, not the same bug." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "SigLIP2 doc example errors are unrelated to local dataset loading in run_image_classification_no_trainer.py." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Nonexistent image_utils import and Qwen3.5 chat crashes are different failures." + }, + { + "left": "issue:44779", + "right": "issue:46489", + "accept": true, + "reason": "Both describe DeepSeek tokenizer regressions in v5; the later issue explicitly frames it as a gap in the prior fix, so they appear to be the same underlying tokenizer bug." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading complaints, but they involve different models and likely different compatibility failures." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Video-input ValueError in Qwen3OmniMoe and KV-cache continuation crash in transformers serve are different code paths." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize/update bug vs Whisper empty-transcription generation bug after align_special_tokens are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:47321", + "accept": false, + "reason": "SigLIP2 inference quality issue and AMD quark class update issue are unrelated." + }, + { + "left": "issue:47145", + "right": "issue:47321", + "accept": false, + "reason": "FP8 weight_scale loading corruption is unrelated to AMD quark class updates." + }, + { + "left": "issue:44242", + "right": "issue:46897", + "accept": false, + "reason": "Load-balancing loss omission in MoE routing and Florence2 double-shift training loss are different bugs." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "trust_remote_code field loss and unrecognized model_type in config.json are different loading/configuration problems." + }, + { + "left": "issue:43673", + "right": "issue:46421", + "accept": false, + "reason": "Both mention chunked prefill/cache, but one is a missing GenerationMixin cache and the other is lazy-init recompilation; not the same concrete bug." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "LoRA/extended-vocab collapse and incorrect visual encoder keys in save_pretrained are unrelated." + }, + { + "left": "issue:45520", + "right": "issue:46170", + "accept": false, + "reason": "flash_attn import on Python 3.13 and Windows ROCm torch.distributed dependency issues are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:47321", + "accept": false, + "reason": "base_model_prefix loading regression and AMD quark class update are unrelated." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "Both are multimodal, but assistant_masks being all zero is a different bug from wrong vision_position_ids in qwen2.5-vl video inputs." + }, + { + "left": "issue:43317", + "right": "issue:47312", + "accept": false, + "reason": "dequantized model offload failure and kernel compatibility issue are unrelated." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "A stale GraniteSpeech CI test and removed Sam3LiteText auto-mappings are different CI failures." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "TensorFlow backend packaging and a generic dependency issue are not the same bug." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "GraniteSpeech CI test failure is unrelated to Phi-4 multimodal model loading." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Import slowdown and unsafe sys.modules access are related at a high level, but not the same concrete failure." + }, + { + "left": "issue:27957", + "right": "issue:35141", + "accept": false, + "reason": "XLMRoberta Flash Attention support and post_init reinitializing output embeddings after resize are unrelated." + }, + { + "left": "issue:44462", + "right": "issue:45907", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are different API issues." + }, + { + "left": "issue:43673", + "right": "issue:46858", + "accept": false, + "reason": "Both concern cache/prefill, but one is a missing cache in v5.0.0 and the other is a device-tensor seq_length causing slice arithmetic failures; not the same bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The cluster is correctly characterized as heterogeneous, the representative issue choice is supported by activity/open-state signals, and the two accepted soft pairs are at least plausibly same-bug matches based on the packet titles. No clear overstatement of evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 43, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:25251", + "issue:29925", + "issue:32090", + "issue:33290", + "issue:35141", + "issue:36246", + "issue:38617", + "issue:39692", + "issue:41628", + "issue:41950", + "issue:42796", + "issue:42994", + "issue:43295", + "issue:43335", + "issue:43404", + "issue:43531", + "issue:43673", + "issue:43976", + "issue:44016", + "issue:44279", + "issue:44336", + "issue:44496", + "issue:44521", + "issue:44683", + "issue:44737", + "issue:44938", + "issue:44964", + "issue:44987", + "issue:45042", + "issue:45083", + "issue:45216", + "issue:45290", + "issue:45335", + "issue:45446", + "issue:45542", + "issue:45593", + "issue:45600", + "issue:45632", + "issue:45759", + "issue:45803", + "issue:46234", + "issue:46424", + "issue:46506", + "issue:46515", + "issue:46752", + "issue:46932", + "issue:47217", + "issue:47333" + ], + "soft_pairs": [ + "issue:32090|issue:35141", + "issue:44964|issue:45600", + "issue:43404|issue:45216", + "issue:42796|issue:44521", + "issue:43531|issue:45600", + "issue:44737|issue:45803", + "issue:44496|issue:47217", + "issue:43335|issue:45593", + "issue:45290|issue:46752", + "issue:45632|issue:46234", + "issue:44496|issue:45042", + "issue:36246|issue:38617", + "issue:25251|issue:29925", + "issue:36246|issue:41628", + "issue:44938|issue:45542", + "issue:43295|issue:45335", + "issue:45083|issue:46506", + "issue:44496|issue:45759", + "issue:41950|issue:44016", + "issue:29925|issue:33290", + "issue:43976|issue:44987", + "issue:39692|issue:46932", + "issue:44683|issue:45446", + "issue:42994|issue:47333", + "issue:44279|issue:46515", + "issue:43673|issue:46424", + "issue:43673|issue:44336" + ], + "prepared_review_unit_hash": "e053a2b815bd1b506a9d3fb8fde16800591d1c9eab20ffa97ef738e42c3f9761", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35866, + "estimated_input_tokens": 8967, + "estimated_eval_tokens": 18190 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a semantic false-positive around assorted Transformers bugs. None of the proposed pairs look like the same concrete bug/change, so I would not merge any soft edge.", + "confidence": 0.69, + "canonical_issue_reason": "Issue #44496 is the most representative anchor in this noisy set: it is a concrete model-loading/configuration failure and has the strongest overlap with the nearby import/load-related reports, though the cluster is too heterogeneous for true deduping.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #44496 is the clearest single user-facing bug to anchor triage in this cluster, but it should be treated as a representative example rather than a duplicate sink for the other issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Different failures and code paths: GPU broadcast/type handling in Trainer vs embedding resize/post-init reinitialization." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Model-loading failure for a specific checkpoint vs CI break from stale auto_mappings references; not the same bug." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Both involve model saving/weight tying, but they are different models and different failure modes, so not one concrete bug." + }, + { + "left": "issue:42796", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer BPE-marker handling is unrelated to the multimodal assistant-mask bug in apply_chat_template." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior and removed-config CI failures are unrelated." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "CPU device-placement bug in XLNet vs bare-except error masking in Fuyu batch conversion; different code paths." + }, + { + "left": "issue:44496", + "right": "issue:47217", + "accept": false, + "reason": "Model/config loading error vs image-placeholder parsing failure in VL text; not the same underlying issue." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "Sparse-layer construction bug vs auxiliary-loss gating in D-FINE; distinct behavior and fix." + }, + { + "left": "issue:45290", + "right": "issue:46752", + "accept": false, + "reason": "Both crash in apply_chat_template, but one is about tool-call assistant messages and the other is an empty-list IndexError; not clearly the same bug." + }, + { + "left": "issue:45632", + "right": "issue:46234", + "accept": false, + "reason": "Cache-path collision for local trust_remote_code models is unrelated to a missing import_utils helper." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Unrecognized model/config issue vs PIL image processors incorrectly requiring torchvision; different failures." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Different missing imports from different modules/symbols; same general symptom, not the same bug." + }, + { + "left": "issue:25251", + "right": "issue:29925", + "accept": false, + "reason": "Pipeline top_k nesting bug and SigLIP use_fast enforcement are unrelated." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but for different symbols and different model families; too broad to merge." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python-version load failure vs TensorFlow/tensorboard backend misdetection; unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Processor/tokenizer regression and resize_token_embeddings not propagating to decoder.embed_tokens are separate bugs." + }, + { + "left": "issue:45083", + "right": "issue:46506", + "accept": false, + "reason": "Helper function length calculation in qwen3_omni_moe vs CPU/CUDA divergence in a different model; not the same bug." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Generic model registration/config failure vs composite multimodal AutoModelForCausalLM unwrapping bug; different code paths." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Video-classification processor lookup and a notebook syntax error are unrelated." + }, + { + "left": "issue:29925", + "right": "issue:33290", + "accept": false, + "reason": "Tokenizer fast-flag enforcement and Deepspeed/Adafactor OOM are not the same issue." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Python 3.9/3.10 compatibility problem vs a specific model-loading failure; not the same change or bug." + }, + { + "left": "issue:39692", + "right": "issue:46932", + "accept": false, + "reason": "SigLIP2 documentation errors and Gemma4 quantized dtype-casting are different model-specific issues." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Compiled flex_attention failing on torch >= 2.9 is distinct from the separate PyTorch version check bug for AuxRequest import." + }, + { + "left": "issue:42994", + "right": "issue:47333", + "accept": false, + "reason": "Quantized model saving and disk-offloaded model saving are related only at a high level; different failure modes and fixes." + }, + { + "left": "issue:44279", + "right": "issue:46515", + "accept": false, + "reason": "Generic dependency issue vs torch 2.4.1 incompatibility; too broad and not the same concrete bug." + }, + { + "left": "issue:43673", + "right": "issue:46424", + "accept": false, + "reason": "GenerationMixin cache missing in chunked_prefill is separate from max_cache_length only applying to Quantized Cache." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Cache/generation behavior vs ANSI formatting in loading_report; unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 44, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:39692", + "issue:41950", + "issue:42796", + "issue:42907", + "issue:42994", + "issue:43295", + "issue:43299", + "issue:43381", + "issue:43441", + "issue:43479", + "issue:43526", + "issue:43577", + "issue:43650", + "issue:43653", + "issue:43824", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44220", + "issue:44279", + "issue:44295", + "issue:44327", + "issue:44466", + "issue:44496", + "issue:44779", + "issue:44792", + "issue:44933", + "issue:45020", + "issue:45216", + "issue:45276", + "issue:45335", + "issue:45375", + "issue:45405", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45812", + "issue:45823", + "issue:46082", + "issue:46121", + "issue:46123", + "issue:46369", + "issue:46424", + "issue:46506", + "issue:46752", + "issue:47217", + "issue:47332" + ], + "soft_pairs": [ + "issue:44295|issue:46506", + "issue:43381|issue:45593", + "issue:43526|issue:43901", + "issue:44327|issue:46752", + "issue:44016|issue:44220", + "issue:46082|issue:46121", + "issue:42994|issue:47332", + "issue:43650|issue:44220", + "issue:44016|issue:44062", + "issue:43441|issue:45588", + "issue:44792|issue:46424", + "issue:42907|issue:43441", + "issue:45335|issue:45584", + "issue:41950|issue:45020", + "issue:43824|issue:44933", + "issue:44279|issue:45405", + "issue:43901|issue:44496", + "issue:43976|issue:44279", + "issue:43295|issue:44779", + "issue:39692|issue:43901", + "issue:43901|issue:47217", + "issue:44779|issue:45812", + "issue:43299|issue:45216", + "issue:42796|issue:43650", + "issue:41950|issue:45823", + "issue:43479|issue:45375", + "issue:44466|issue:45276", + "issue:43577|issue:45216", + "issue:46123|issue:46369", + "issue:43653|issue:45593" + ], + "prepared_review_unit_hash": "23a28430c7f36e36dfb872e240c9e3f50636abd46ca0026f2118f479053f9355", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36672, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated Transformers regressions, docs issues, tokenizer/model bugs, and save/load edge cases. I did not find any soft pair that looks like the same underlying bug/change, so no duplicate merges are justified.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 45020 is the broadest and most central regression report in the set, but the items here are not true duplicates; it is only the best umbrella-style representative if one must be chosen.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the strongest standalone issue to keep because it covers a wide, user-impacting regression class (recent versions breaking remote-code models) and is more general than the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44295", + "right": "issue:46506", + "accept": false, + "reason": "Different bugs: one is a position_ids buffer-read error, the other is a CPU/CUDA output divergence in PPDocLayoutV3." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Different code paths and symptoms: eval/gradient-checkpointing conflict vs D-FINE auxiliary-loss handling when denoising is off." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "One is a BeitImageProcessorFast label-reduction bug; the other is a docs/behavior mismatch for TextClassificationPipeline." + }, + { + "left": "issue:44327", + "right": "issue:46752", + "accept": false, + "reason": "Different pipelines and failures: QA decode_spans top-k bounds error vs empty-list apply_chat_template crash." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "A notebook syntax error is unrelated to the _torch_extract_fbank_features runtime issue." + }, + { + "left": "issue:46082", + "right": "issue:46121", + "accept": false, + "reason": "These are separate validation/config bugs: explicit head_dim rejection vs JSON-loaded list handling in rope validation." + }, + { + "left": "issue:42994", + "right": "issue:47332", + "accept": false, + "reason": "Both involve saving, but one is quantized-model saving and the other is disk-offloaded model saving; not the same failure." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "The placeholder title is unrelated to the audio feature extraction bug." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error and AddedToken keyword collision are unrelated." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Both mention FlashAttention, but one is a Ministral-3 v5 RC failure and the other is an s_aux=None AttributeError in sink-less models." + }, + { + "left": "issue:44792", + "right": "issue:46424", + "accept": false, + "reason": "A janus image-generation test failure is unrelated to cache max length being honored only for Quantized Cache." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Both are model-specific failures, but one is save-dequantized Ministrals/Devstrals and the other is a FlashAttention issue." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "Different models and mechanics: token embedding resize propagation vs Whisper empty-transcription generation after align_special_tokens." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "A video-classification processor lookup bug is not the same as the broad remote_code regression." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Both are import errors, but for different symbols/modules and likely different root causes." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Generic dependency trouble is not the same as a PEFT minimum-version bump to an unreleased release." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "Docs drift for return_all_scores is unrelated to unrecognized model/config.json issues." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "Python version compatibility and dependency issues are too different to treat as duplicates." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Different regressions: processor/tokenizer custom-code breakage vs Deepseek tokenizer output corruption." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Both touch docs, but SigLIP2 example errors and TextClassificationPipeline docs drift are unrelated." + }, + { + "left": "issue:43901", + "right": "issue:47217", + "accept": false, + "reason": "Docs mismatch vs VL image-placeholder runtime failure are not the same bug." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and likely different tokenizer code paths." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Model loading failure and checkpoint-saving regression are different operations with different symptoms." + }, + { + "left": "issue:42796", + "right": "issue:43650", + "accept": false, + "reason": "Tokenizer BPE-marker handling is unrelated to the placeholder 'ADD THE DATA' issue." + }, + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Video-classification processor lookup and Gemma4 PLE device mismatch are unrelated bugs." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Different multimodal config bugs for different models/fields; no shared concrete code path is evident." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Serialization of tied lm_head weights and Gemma4 resize_token_embeddings propagation are separate issues." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "BLIP2 dtype loading behavior is unrelated to Qwen3.5 save_pretrained checkpoint corruption." + }, + { + "left": "issue:46123", + "right": "issue:46369", + "accept": false, + "reason": "A partial-batch mask pipeline bug is unrelated to sam3-litetext memory overflow." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and D-FINE auxiliary-loss behavior are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 45, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:42994", + "issue:43299", + "issue:43317", + "issue:43381", + "issue:43441", + "issue:43644", + "issue:43650", + "issue:43854", + "issue:43873", + "issue:43931", + "issue:43976", + "issue:44016", + "issue:44075", + "issue:44242", + "issue:44246", + "issue:44291", + "issue:44336", + "issue:44351", + "issue:44368", + "issue:44462", + "issue:44671", + "issue:44683", + "issue:44779", + "issue:45081", + "issue:45103", + "issue:45200", + "issue:45278", + "issue:45357", + "issue:45372", + "issue:45405", + "issue:45542", + "issue:45588", + "issue:45800", + "issue:45823", + "issue:45834", + "issue:45859", + "issue:45907", + "issue:46082", + "issue:46209", + "issue:46399", + "issue:46512", + "issue:46614", + "issue:46851", + "issue:47000", + "issue:47085", + "issue:47145", + "issue:47217", + "issue:47405" + ], + "soft_pairs": [ + "issue:46614|issue:47085", + "issue:47145|issue:47405", + "issue:43976|issue:45405", + "issue:43381|issue:47000", + "issue:44462|issue:45081", + "issue:43644|issue:47145", + "issue:44075|issue:44368", + "issue:43299|issue:45357", + "issue:46082|issue:46399", + "issue:43299|issue:43854", + "issue:43931|issue:47145", + "issue:44242|issue:45200", + "issue:44683|issue:45542", + "issue:46082|issue:46512", + "issue:44016|issue:44291", + "issue:45588|issue:47000", + "issue:44351|issue:45278", + "issue:43650|issue:46209", + "issue:43873|issue:47145", + "issue:45859|issue:46851", + "issue:45081|issue:45907", + "issue:45103|issue:45372", + "issue:42994|issue:43317", + "issue:45542|issue:45800", + "issue:43441|issue:45834", + "issue:44336|issue:46851", + "issue:44246|issue:47217", + "issue:44671|issue:44779", + "issue:44016|issue:45823" + ], + "prepared_review_unit_hash": "81800b54c0ce08bc5f411f3ce16c00a8f725884e116ea8a12c9cbb355f2eecce", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36413, + "estimated_input_tokens": 9104, + "estimated_eval_tokens": 18464 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose semantic grouping of mostly unrelated Transformers v5 regressions and model-specific bugs, centered around loading, quantization/offload, tokenizers, and a few unrelated utilities. None of the soft pairs look like true duplicates.", + "confidence": 0.92, + "canonical_issue_reason": "issue:47405 is the best anchor for the cluster because it describes a high-severity, broadly framed model-loading correctness regression (silent wrong weights from an Auto-class mismatch) and is more representative than the many niche, subsystem-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:47405 is the strongest single issue to represent the group: it is clearly actionable, severe, and closest to the cluster\u2019s shared theme of loading/serialization regressions. The rest are mostly separate bugs rather than duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46614", + "right": "issue:47085", + "accept": false, + "reason": "Different components and failure modes: stale phonemizer backend state vs tokenizer thread-safety regression." + }, + { + "left": "issue:47145", + "right": "issue:47405", + "accept": false, + "reason": "Both are silent-corruption-style bugs, but one is FP8 weight-scale loading and the other is wrong Auto-class conversion/loading; different code paths." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "One is a Python version compatibility bug; the other is an unreleased dependency version bump." + }, + { + "left": "issue:43381", + "right": "issue:47000", + "accept": false, + "reason": "Unrelated bugs: gradient checkpointing in eval mode vs RTDetr loss infeasible cost matrix." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer-file precedence bug vs Mistral regex patch crash; different loading paths." + }, + { + "left": "issue:43644", + "right": "issue:47145", + "accept": false, + "reason": "Non-persistent buffer initialization bug vs FP8 compressed-tensor weight-scale loading bug." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "Optimizer argument handling vs a model warning about tie_word_embeddings; not the same defect." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Different model families and symptoms: Qwen3VL MoE loading breakage vs Qwen3.5 save_pretrained visual-key regression." + }, + { + "left": "issue:46082", + "right": "issue:46399", + "accept": false, + "reason": "Config validation for head_dim vs a kernel registration/decorator crash; separate issues." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Both are model-loading problems, but for different models and distinct failure mechanisms." + }, + { + "left": "issue:43931", + "right": "issue:47145", + "accept": false, + "reason": "Shape-mismatch loading failure vs FP8 checkpoint scaling corruption; different underlying bugs." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "MoE load-balancing loss logic vs Gemma 4 multimodal token-type defaults." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Torch/compiled attention incompatibility vs TensorFlow/tensorboard backend import issue." + }, + { + "left": "issue:46082", + "right": "issue:46512", + "accept": false, + "reason": "Different API contracts: LlamaConfig head_dim validation vs from_config dtype/config synchronization." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error vs init_empty_weights argument mismatch; no shared code path." + }, + { + "left": "issue:45588", + "right": "issue:47000", + "accept": false, + "reason": "FlashAttention sink handling crash vs RTDetrLoss infeasible cost matrix are unrelated." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "Specific missing HybridCache import vs broad import errors after upgrade; not clearly the same bug." + }, + { + "left": "issue:43650", + "right": "issue:46209", + "accept": false, + "reason": "Placeholder issue with no bug detail vs FP8 inference failure on B200." + }, + { + "left": "issue:43873", + "right": "issue:47145", + "accept": false, + "reason": "Quantized offloading behavior vs missing weight_scale application; too different to merge." + }, + { + "left": "issue:45859", + "right": "issue:46851", + "accept": false, + "reason": "CPU-offload incompatibility in rotary embedding vs variable-length position_ids handling." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Mistral tokenizer regex patch crash vs list_repo_files revision forwarding; unrelated utilities." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "Auto-docstring annotation parsing crash vs processor loading broken by ReasoningEffort import." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Quantized model saving failure vs dequantized model loading with device_map=auto; related theme, but different defect." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "TensorFlow backend detection bug vs torch/transformers version incompatibility; not duplicates." + }, + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "FlashAttention failure on Ministral-3 vs OCR long-input index error in Kosmos2.5; different models and failures." + }, + { + "left": "issue:44336", + "right": "issue:46851", + "accept": false, + "reason": "ANSI output formatting issue vs Qwen3.5 variable-length position_ids bug." + }, + { + "left": "issue:44246", + "right": "issue:47217", + "accept": false, + "reason": "Import-time slowdown vs VL image-placeholder handling bug; unrelated." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizer regressions in different models: CamemBERT masked LM vs Deepseek tokenizer outputs." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error vs Gemma4 device mismatch during forward; different layers of the stack." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 46, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:33290", + "issue:42175", + "issue:42617", + "issue:42796", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43575", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43784", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:44162", + "issue:44186", + "issue:44188", + "issue:44246", + "issue:44261", + "issue:44263", + "issue:44297", + "issue:44336", + "issue:44373", + "issue:44496", + "issue:44589", + "issue:44623", + "issue:44749", + "issue:45325", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45588", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:46291", + "issue:46399", + "issue:46655", + "issue:46726", + "issue:46846", + "issue:46851", + "issue:47000", + "issue:47145", + "issue:47333", + "issue:47427" + ], + "soft_pairs": [ + "issue:43784|issue:45542", + "issue:44186|issue:47000", + "issue:42175|issue:43976", + "issue:44623|issue:44749", + "issue:45325|issue:46851", + "issue:44623|issue:45397", + "issue:43673|issue:43761", + "issue:43901|issue:44246", + "issue:43525|issue:43881", + "issue:43479|issue:43525", + "issue:33290|issue:45656", + "issue:43575|issue:47145", + "issue:46399|issue:46846", + "issue:44589|issue:45362", + "issue:44261|issue:44263", + "issue:44188|issue:44336", + "issue:42796|issue:43441", + "issue:45600|issue:45712", + "issue:45588|issue:47427", + "issue:45362|issue:47333", + "issue:43824|issue:43976", + "issue:44246|issue:44496", + "issue:44336|issue:44373", + "issue:43723|issue:44162", + "issue:42617|issue:43976", + "issue:46291|issue:46399", + "issue:44297|issue:45397", + "issue:46399|issue:46851", + "issue:46655|issue:46726" + ], + "prepared_review_unit_hash": "7e18e5938c559d28886a9463be5b1e739adb29f52b879f09832c0ba124b02d39", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35756, + "estimated_input_tokens": 8939, + "estimated_eval_tokens": 18134 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is a grab bag of unrelated Transformers issues rather than a tight duplicate set. The soft-similarity pairs are mostly false positives: they share broad subsystems or themes, but not the same concrete bug or change path. No PRs were provided.", + "confidence": 0.87, + "canonical_issue_reason": "issue:43824 is the strongest standalone anchor: it has a concrete import failure, multiple inbound references, and substantial discussion, making it the most actionable representative issue in an otherwise heterogeneous set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43824 is the best global issue candidate because it describes a specific, user-facing regression with clear reproduction and ecosystem impact, unlike many of the other items that are narrower or orthogonal.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Both are import/backend failures, but they are different bugs: sentence-transformers import hits a missing nn symbol, while tensorboard-only installs trigger a missing TensorFlow backend path." + }, + { + "left": "issue:44186", + "right": "issue:47000", + "accept": false, + "reason": "Completely different failures: a tokenizer crash on NER/padding versus an RTDetrLoss infeasible cost-matrix error." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "One is a packaging/backend omission for pip extras, the other is a Python version compatibility problem; not the same underlying issue." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Saving processor files and post-upgrade performance regression are unrelated." + }, + { + "left": "issue:45325", + "right": "issue:46851", + "accept": false, + "reason": "Both mention position_ids, but they affect different Qwen models and different code paths: still-image temporal scaling vs variable-length input handling." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Processor save-pretrained file omission is unrelated to a Zero3 from_pretrained loading failure." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Generation cache handling during chunked prefill is unrelated to CLIPVisionModel hidden_states being omitted." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Docs mismatch for return_all_scores is not the same problem as intermittent slow imports." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config and a glm-4v-9b load failure are different model-specific bugs." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Default multimodal config initialization and missing pad_token_id are separate config issues." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Both involve deepspeed/optimizer behavior, but one is OOM with Adafactor and the other is optimizer.step() being invoked twice; not mergeable as one fix." + }, + { + "left": "issue:43575", + "right": "issue:47145", + "accept": false, + "reason": "Tensor-parallel OOM on Qwen2-57B is unrelated to silent FP8 checkpoint corruption and missing weight_scale application." + }, + { + "left": "issue:46399", + "right": "issue:46846", + "accept": false, + "reason": "Different Qwen3.5 failures: kernel registration/decorator error versus missing TP-plan entries causing OOM/shape errors." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "A storage-object lookup TypeError is unrelated to a chat crash on Qwen3.5." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Different model internals and failures: MLA normalization precision/config issue versus torch.split return handling." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Attention-kernel divergence under torch.compile is unrelated to ANSI codes in loading_report." + }, + { + "left": "issue:42796", + "right": "issue:43441", + "accept": false, + "reason": "Tokenizer BPE-marker handling and FlashAttention support are different bugs, even though both involve Ministral-3." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Removed auto-mapping references breaking CI and leftover dummy PT classes are separate cleanup problems." + }, + { + "left": "issue:45588", + "right": "issue:47427", + "accept": false, + "reason": "Flash-attention sink handling and redundant HF init on non-CUDA hardware are different code paths." + }, + { + "left": "issue:45362", + "right": "issue:47333", + "accept": false, + "reason": "Qwen3.5 chat crash is unrelated to saving a disk-offloaded Qwen3-30B-A3B model." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "An import error for a specific model class is unrelated to a Python version support issue." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Intermittent import slowness and unrecognized-model config parsing are not the same bug." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "ANSI output leakage and a wrong docstring are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading in v5 and ESM2 breakage are different failures with no shared concrete code path." + }, + { + "left": "issue:42617", + "right": "issue:43976", + "accept": false, + "reason": "Running 3d_parallel.py failing and Python 3.9 compatibility are unrelated." + }, + { + "left": "issue:46291", + "right": "issue:46399", + "accept": false, + "reason": "kernels package metadata requirements and a missing decorator in Qwen3.5 kernel registration are different issues." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to gemma-4 Zero3 loading." + }, + { + "left": "issue:46399", + "right": "issue:46851", + "accept": false, + "reason": "Kernel registration failure and variable-length position_ids handling are separate Qwen3.5 problems." + }, + { + "left": "issue:46655", + "right": "issue:46726", + "accept": false, + "reason": "Video padding calculation and missing vision-backend image processor resolution are different processor bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 47, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:36246", + "issue:41950", + "issue:42491", + "issue:42915", + "issue:43257", + "issue:43299", + "issue:43526", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44336", + "issue:44479", + "issue:44496", + "issue:44779", + "issue:44871", + "issue:45003", + "issue:45072", + "issue:45083", + "issue:45103", + "issue:45137", + "issue:45292", + "issue:45375", + "issue:45397", + "issue:45561", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:45701", + "issue:45759", + "issue:46399", + "issue:46424", + "issue:46506", + "issue:46614", + "issue:46655", + "issue:46665", + "issue:46726", + "issue:46752", + "issue:46821", + "issue:46829", + "issue:46851", + "issue:47176", + "issue:47177", + "issue:47217", + "issue:47312" + ], + "soft_pairs": [ + "issue:45072|issue:46829", + "issue:42491|issue:43299", + "issue:45083|issue:46851", + "issue:47176|issue:47177", + "issue:45003|issue:47217", + "issue:43526|issue:45003", + "issue:45588|issue:46851", + "issue:45003|issue:46821", + "issue:44246|issue:44871", + "issue:45103|issue:46399", + "issue:44295|issue:45397", + "issue:45561|issue:45563", + "issue:44336|issue:46424", + "issue:43550|issue:46506", + "issue:43540|issue:46752", + "issue:36246|issue:43596", + "issue:46614|issue:46665", + "issue:42915|issue:47312", + "issue:43540|issue:46655", + "issue:46726|issue:47217", + "issue:43257|issue:45137", + "issue:44779|issue:45292", + "issue:44479|issue:46655", + "issue:44496|issue:45003", + "issue:41950|issue:44220", + "issue:45375|issue:45759", + "issue:45292|issue:45701", + "issue:44242|issue:45593" + ], + "prepared_review_unit_hash": "e639f8d3dc4ea1331d995e396bc0b022f965ce8853b43758027b45942a92b19e", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36177, + "estimated_input_tokens": 9045, + "estimated_eval_tokens": 18346 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: a few pairs share model family or subsystem vocabulary, but none look like the same underlying bug/change closely enough to merge as duplicates. No PRs are present.", + "confidence": 0.9, + "canonical_issue_reason": "No single issue is a safe canonical duplicate target; the set mixes unrelated bugs across Qwen/VL, distributed loading, tokenizer/config, import-time, security, and utility code paths.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45003 is the closest broad/core-library candidate and appears in multiple soft-similarity links, but it still does not represent the rest of the cluster as a duplicate.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45072", + "right": "issue:46829", + "accept": false, + "reason": "Different failures: one is a bfloat16 dtype mismatch in model inference, the other is an output field/hidden_states mix-up in VideoPrism classification." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen3 MoE loading, but one is LoRA compatibility across versions and the other is accelerate/deepspeed weight conversion; not the same bug." + }, + { + "left": "issue:45083", + "right": "issue:46851", + "accept": false, + "reason": "Both involve Qwen3.5 positional/length behavior, but one is a helper output-length issue and the other is variable-length position_ids handling." + }, + { + "left": "issue:47176", + "right": "issue:47177", + "accept": false, + "reason": "Distinct security issues: shell command-substitution bypass vs tar/zip-slip arbitrary file write." + }, + { + "left": "issue:45003", + "right": "issue:47217", + "accept": false, + "reason": "sys.modules access in modeling_utils is unrelated to VL image placeholder handling in user text." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BEiT image-processor label reduction and unsafe sys.modules access are unrelated code paths." + }, + { + "left": "issue:45588", + "right": "issue:46851", + "accept": false, + "reason": "Flash-attention s_aux=None crash is a multimodal attention bug, not the same as Qwen3.5 variable-length position_ids handling." + }, + { + "left": "issue:45003", + "right": "issue:46821", + "accept": false, + "reason": "Unsafe sys.modules access and TimesFM window_size AttributeError are different core-library bugs." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Load-balancing loss omission and eos_token_id config inconsistency affect different model behaviors and code paths." + }, + { + "left": "issue:45103", + "right": "issue:46399", + "accept": false, + "reason": "Auto-docstring future-annotations crash is unrelated to the Qwen3.5 kernel registration error." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Reading position_ids after buffer registration and gemma-4 ZeRO3 from_pretrained failures are different distributed/model-loading problems." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "pytest-xdist file race and stale num_return_sequences warning are unrelated testing/generate issues." + }, + { + "left": "issue:44336", + "right": "issue:46424", + "accept": false, + "reason": "ANSI leakage in loading_report and cache max_length behavior are unrelated utility vs cache-config bugs." + }, + { + "left": "issue:43550", + "right": "issue:46506", + "accept": false, + "reason": "torch.compile SDPA failure in Bamba and CPU/CUDA divergence in PPDocLayoutV3 are different backend/model issues." + }, + { + "left": "issue:43540", + "right": "issue:46752", + "accept": false, + "reason": "Qwen3OmniMoe video-input ValueError and empty-list chat template crash are unrelated failures." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "ImportError for Qwen2_5_VLImageProcessor and DeepSpeed ZeRO3 BertModel IndexError are different bugs." + }, + { + "left": "issue:46614", + "right": "issue:46665", + "accept": false, + "reason": "Stale phonemizer backend state and from_pretrained behavior inside Python classes do not describe the same issue." + }, + { + "left": "issue:42915", + "right": "issue:47312", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and kernels 0.16 incompatibility are different compatibility problems." + }, + { + "left": "issue:43540", + "right": "issue:46655", + "accept": false, + "reason": "Both are video-related, but Qwen3OmniMoe ValueError and SmolVLM padding miscalculation are distinct model-specific bugs." + }, + { + "left": "issue:46726", + "right": "issue:47217", + "accept": false, + "reason": "Missing vision backend / unrecognized image processor is not the same as failing when the image placeholder appears in user text." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "MoE weight conversion with accelerate+deepspeed and ZeRO3 empty-deque pop are different distributed-loading failures." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Tokenizer changes across versions and resize_token_embeddings not updating output_embeddings are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:46655", + "accept": false, + "reason": "General v5.3.0 video-input regression across several VL models and SmolVLM padding calculation are related only at a very broad level, not the same concrete bug." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model/model_type config error and unsafe sys.modules access are unrelated loading-time issues." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video-classification pipeline choosing image processors and _torch_extract_fbank_features() are different pipeline defects." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Missing deepstack_visual_indexes field and composite Qwen multimodal text_config unwrapping are distinct config-handling bugs." + }, + { + "left": "issue:45292", + "right": "issue:45701", + "accept": false, + "reason": "resize_token_embeddings not affecting output_embeddings and version-dependent tokenization changes are different behaviors." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Load-balancing loss gated on output_router_logits and D-FINE auxiliary-loss omission with denoising off are separate model-loss bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 48, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:29870", + "issue:30333", + "issue:36246", + "issue:42757", + "issue:42886", + "issue:43295", + "issue:43352", + "issue:43475", + "issue:43540", + "issue:44038", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44295", + "issue:44351", + "issue:44451", + "issue:44466", + "issue:44479", + "issue:44623", + "issue:44749", + "issue:44877", + "issue:44987", + "issue:44995", + "issue:45003", + "issue:45137", + "issue:45200", + "issue:45250", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45397", + "issue:45459", + "issue:45561", + "issue:45588", + "issue:45600", + "issue:45656", + "issue:45685", + "issue:46132", + "issue:46170", + "issue:46369", + "issue:46431", + "issue:46506", + "issue:46726", + "issue:46752", + "issue:47217", + "issue:47321", + "issue:47429" + ], + "soft_pairs": [ + "issue:43295|issue:45292", + "issue:43352|issue:45250", + "issue:44995|issue:45200", + "issue:36246|issue:44351", + "issue:36246|issue:42757", + "issue:44060|issue:44077", + "issue:45137|issue:45600", + "issue:46170|issue:47321", + "issue:44466|issue:45335", + "issue:44162|issue:45200", + "issue:42886|issue:46132", + "issue:44295|issue:44749", + "issue:44995|issue:45588", + "issue:44451|issue:46752", + "issue:45685|issue:46431", + "issue:44877|issue:45375", + "issue:44479|issue:46170", + "issue:47217|issue:47429", + "issue:43475|issue:46369", + "issue:45003|issue:45561", + "issue:44877|issue:46170", + "issue:44623|issue:46726", + "issue:44987|issue:45459", + "issue:44038|issue:46506", + "issue:29870|issue:30333", + "issue:45397|issue:45656", + "issue:43475|issue:43540" + ], + "prepared_review_unit_hash": "a57951c26c2be290aefedabcd3aebc240fc15a01ecef0310828faac39a2ca9ba", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35126, + "estimated_input_tokens": 8782, + "estimated_eval_tokens": 17820 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is not a true duplicate cluster: it contains many unrelated bug reports, with only superficial overlap in words like \"tokenizer\", \"video\", \"config\", or \"deepspeed\". None of the soft pairs look like the same underlying bug/change, so no dedupes should be merged.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue is appropriate here; the items are heterogeneous and do not collapse to one underlying bug.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue: selecting one would be arbitrary because the cluster is not coherent.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "Both mention embedding/processor regressions, but one is about custom multimodal processor.tokenizer behavior and the other about resize_token_embeddings not updating output embeddings; different code paths and failure modes." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "43352 is a specific model gate for NemotronHForCausalLM, while 45250 is just a general Flash Attention topic; not the same bug." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "Stale indexer cache crash on second forward is unrelated to Gemma 4 requiring mm_token_type_ids to default to zeros." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different missing symbols from different modules and versions; not the same underlying issue." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Qwen2_5_VLImageProcessor import failure is unrelated to huggingface_hub.is_offline_mode missing." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "One is a tied-weights warning bug in Qwen3-Next; the other is a model policy change for patchtsmixer post_init. Different areas and fixes." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque underflow and auto_mappings referencing removed Sam3LiteText configs are unrelated." + }, + { + "left": "issue:46170", + "right": "issue:47321", + "accept": false, + "reason": "Windows ROCm/torch.distributed import problems are unrelated to AMD quark class updates." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Both involve resizing/tied embeddings, but one is serialization of lm_head.weight and the other is decoder.embed_tokens not updating; distinct bugs." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids behavior are completely different models and code paths." + }, + { + "left": "issue:42886", + "right": "issue:46132", + "accept": false, + "reason": "Tokenizer cache loading in offline mode is unrelated to AttentionInterface.register changing function behavior." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "Reading position_ids after buffer registration and a large slowdown after upgrading transformers are different regressions." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Stale indexer cache on second forward is unrelated to flash_attention s_aux=None AttributeError." + }, + { + "left": "issue:44451", + "right": "issue:46752", + "accept": false, + "reason": "Failed model load for ScandiBERT and apply_chat_template crashing on empty input are unrelated." + }, + { + "left": "issue:45685", + "right": "issue:46431", + "accept": false, + "reason": "MPS histogram Int error in moe code is unrelated to missing chat template on a processor." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both mention strict config handling, but one blocks loading granite_speech and the other drops a missing field in Qwen3_5MoeVisionConfig; not the same artifact or fix." + }, + { + "left": "issue:44479", + "right": "issue:46170", + "accept": false, + "reason": "Qwen video-input regression and Windows ROCm import/type errors are unrelated." + }, + { + "left": "issue:47217", + "right": "issue:47429", + "accept": false, + "reason": "VL image-placeholder parsing and a tokenizers version upper-bound issue are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:46369", + "accept": false, + "reason": "Both are SAM3-family, but one is a missing attribute in video vision encoder output and the other is a memory overflow in sam3-litetext; not the same bug." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Unsafe sys.modules access and pytest-xdist race on captured_info.txt are different concurrency-related problems." + }, + { + "left": "issue:44877", + "right": "issue:46170", + "accept": false, + "reason": "granite_speech strict config loading and Windows ROCm torch.distributed import issues are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:46726", + "accept": false, + "reason": "Missing files from processor.save_pretrained and unrecognized image processor when a vision backend is missing are different failure modes." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Model loading failure for physical-intelligence/fast and protobuf-related tokenizer error masking are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:46506", + "accept": false, + "reason": "Qwen3-VL-Moe bug and CPU/CUDA result divergence in PPDocLayoutV3 are different models and code paths." + }, + { + "left": "issue:29870", + "right": "issue:30333", + "accept": false, + "reason": "Bart generate UnboundLocalError and MLflow job status reporting are unrelated." + }, + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Gemma-4 ZeRO3 from_pretrained issues and a double optimizer step with deepspeed are separate bugs." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "Both are multimodal/video-related, but one is a missing attribute in SAM3 output and the other is a Qwen3OmniMoe video-processing ValueError; different models and fixes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 49, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:42199", + "issue:42886", + "issue:42994", + "issue:43065", + "issue:43408", + "issue:43653", + "issue:43701", + "issue:43723", + "issue:43784", + "issue:43906", + "issue:43931", + "issue:43976", + "issue:43994", + "issue:44162", + "issue:44188", + "issue:44242", + "issue:44361", + "issue:44368", + "issue:44483", + "issue:44749", + "issue:44857", + "issue:44977", + "issue:45103", + "issue:45229", + "issue:45412", + "issue:45563", + "issue:45593", + "issue:45656", + "issue:45987", + "issue:46082", + "issue:46123", + "issue:46132", + "issue:46202", + "issue:46305", + "issue:46315", + "issue:46399", + "issue:46421", + "issue:46710", + "issue:46726", + "issue:46752", + "issue:46851", + "issue:46858", + "issue:47020", + "issue:47059", + "issue:47068", + "issue:47145", + "issue:47312", + "issue:47429" + ], + "soft_pairs": [ + "issue:44242|issue:44857", + "issue:44368|issue:44977", + "issue:43653|issue:44162", + "issue:46305|issue:46315", + "issue:42994|issue:47312", + "issue:43065|issue:43408", + "issue:43723|issue:46710", + "issue:45412|issue:46123", + "issue:42886|issue:45229", + "issue:47059|issue:47429", + "issue:43931|issue:44368", + "issue:43906|issue:46202", + "issue:45987|issue:46082", + "issue:45103|issue:46858", + "issue:44361|issue:44749", + "issue:42199|issue:46851", + "issue:43994|issue:46726", + "issue:44368|issue:45563", + "issue:46132|issue:47145", + "issue:46752|issue:47020", + "issue:44188|issue:46202", + "issue:45593|issue:47068", + "issue:43976|issue:47429", + "issue:43701|issue:44749", + "issue:44483|issue:46202", + "issue:43784|issue:46399", + "issue:44188|issue:46421", + "issue:44749|issue:45656" + ], + "prepared_review_unit_hash": "7fa01b1d0b4596fde07b06a4579234fc3a78ebfeb72346797d21a4073e1f134c", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36278, + "estimated_input_tokens": 9070, + "estimated_eval_tokens": 18396 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These issues are largely unrelated; none of the soft pairs look like the same underlying bug or change, so I rejected all candidate duplicates. The strongest standalone issue in the set is 45229 because it has the richest discussion and the clearest bug context.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45229 is the most complete and high-signal bug report in the set, with extensive discussion and a concrete multi-GPU CUDA OOM reproduction.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45229 is the best overall issue candidate because it is specific, well-documented, and has the most diagnostic context.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Different problems: MoE load-balancing loss vs D-FINE image loss/AMP crash." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated Qwen3.5 warning about tie_word_embeddings vs flash-attention generation failure." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token bug is unrelated to ESM2 model breakage." + }, + { + "left": "issue:46305", + "right": "issue:46315", + "accept": false, + "reason": "Sliding-window mask off-by-one is unrelated to tokenizer conversion slowdown." + }, + { + "left": "issue:42994", + "right": "issue:47312", + "accept": false, + "reason": "Quantized model saving failure is unrelated to kernel version incompatibility." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Both touch SAM3, but one is a dummy Conv2d concern and the other is a model-type warning; not the same bug." + }, + { + "left": "issue:43723", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer loading in v5 is unrelated to an accuracy regression in DeepSeek-R1." + }, + { + "left": "issue:45412", + "right": "issue:46123", + "accept": false, + "reason": "Memory not released in RT-DETR is unrelated to partial-batch result dropping in MaskGenerationPipeline." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading failure is unrelated to Gemma4 multi-GPU OOM." + }, + { + "left": "issue:47059", + "right": "issue:47429", + "accept": false, + "reason": "CLI completion breakage with huggingface_hub is unrelated to tokenizers version pinning." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch is unrelated to a tie_word_embeddings warning." + }, + { + "left": "issue:43906", + "right": "issue:46202", + "accept": false, + "reason": "Isolated reproduction of another issue is not the same as Gemma pipeline roundtrip failure on \u2581." + }, + { + "left": "issue:45987", + "right": "issue:46082", + "accept": false, + "reason": "StaticCache seq-length contract issue is unrelated to LlamaConfig head_dim validation." + }, + { + "left": "issue:45103", + "right": "issue:46858", + "accept": false, + "reason": "auto_docstring annotations crash is unrelated to static-cache seq-length device-tensor arithmetic." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "MLukeTokenizer task AttributeError is unrelated to a training slowdown regression." + }, + { + "left": "issue:42199", + "right": "issue:46851", + "accept": false, + "reason": "DETR cardinality error and Qwen3.5 variable-length position_ids are different code paths." + }, + { + "left": "issue:43994", + "right": "issue:46726", + "accept": false, + "reason": "SigLIP2 nonsensical outputs are unrelated to AutoProcessor missing-backend handling." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Different warning issues: tie_word_embeddings vs stale num_return_sequences warning." + }, + { + "left": "issue:46132", + "right": "issue:47145", + "accept": false, + "reason": "AttentionInterface.register behavior change is unrelated to FP8 checkpoint scale application." + }, + { + "left": "issue:46752", + "right": "issue:47020", + "accept": false, + "reason": "Empty chat template list crash is unrelated to XLMRobertaTokenizer Unigram init mismatch." + }, + { + "left": "issue:44188", + "right": "issue:46202", + "accept": false, + "reason": "torch.compile attention-kernel branching bug is unrelated to Gemma pipeline token roundtrip failure." + }, + { + "left": "issue:45593", + "right": "issue:47068", + "accept": false, + "reason": "Missing auxiliary losses in D-FINE is unrelated to deepspeed_sp_compute_loss host synchronization." + }, + { + "left": "issue:43976", + "right": "issue:47429", + "accept": false, + "reason": "Python version compatibility regression is unrelated to tokenizers upper-bound packaging issue." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "resume_from_checkpoint key mismatch is unrelated to the training slowdown regression." + }, + { + "left": "issue:44483", + "right": "issue:46202", + "accept": false, + "reason": "Chat completions API rejection is unrelated to Gemma pipeline roundtrip token handling." + }, + { + "left": "issue:43784", + "right": "issue:46399", + "accept": false, + "reason": "Import-time NameError in sentence-transformers is unrelated to Qwen3.5 kernel registration failure." + }, + { + "left": "issue:44188", + "right": "issue:46421", + "accept": false, + "reason": "Both involve compile/cache behavior, but they describe different failure modes and code paths." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Training slowdown after upgrade is unrelated to optimizer step being called twice under deepspeed." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 50, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:36010", + "issue:39692", + "issue:41628", + "issue:42994", + "issue:43065", + "issue:43232", + "issue:43475", + "issue:43482", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43723", + "issue:43873", + "issue:44038", + "issue:44162", + "issue:44188", + "issue:44279", + "issue:44297", + "issue:44373", + "issue:44451", + "issue:44623", + "issue:44661", + "issue:44869", + "issue:44871", + "issue:44945", + "issue:44964", + "issue:45137", + "issue:45230", + "issue:45356", + "issue:45375", + "issue:45397", + "issue:45478", + "issue:45561", + "issue:45656", + "issue:46123", + "issue:46132", + "issue:46326", + "issue:46463", + "issue:46726", + "issue:46821", + "issue:46851", + "issue:46877", + "issue:47000", + "issue:47059", + "issue:47217", + "issue:47429" + ], + "soft_pairs": [ + "issue:44297|issue:44623", + "issue:39692|issue:46726", + "issue:43065|issue:46123", + "issue:43475|issue:46821", + "issue:44964|issue:45137", + "issue:43475|issue:43643", + "issue:42994|issue:43482", + "issue:43646|issue:44162", + "issue:46463|issue:46877", + "issue:45230|issue:45397", + "issue:44373|issue:46851", + "issue:44623|issue:45656", + "issue:28282|issue:30990", + "issue:44871|issue:46132", + "issue:43723|issue:46326", + "issue:44188|issue:46851", + "issue:46877|issue:47000", + "issue:45561|issue:47217", + "issue:43232|issue:43673", + "issue:44451|issue:47429", + "issue:44279|issue:47429", + "issue:44038|issue:44623", + "issue:44869|issue:45356", + "issue:36010|issue:41628", + "issue:45375|issue:45478", + "issue:44661|issue:47059", + "issue:43873|issue:44945" + ], + "prepared_review_unit_hash": "0f7dda074c9894dadba4425d8a8b807e5f66c5a6a395756f617e1d5f042633e6", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35373, + "estimated_input_tokens": 8844, + "estimated_eval_tokens": 17944 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose collection of mostly unrelated Transformers regressions and compatibility bugs, with a few broad v5-era breakages around loading, tokenizers/processors, generation, and quantization. I do not see any soft-pair duplicates that look mergeable into one fix.", + "confidence": 0.57, + "canonical_issue_reason": "issue:43646 is the best umbrella candidate: it is broad, v5-specific, and closer to the recurring theme of core model initialization regressions than the more niche or model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the most representative single issue in the set because it captures a central Transformers 5.0 breakage rather than a narrow model- or docs-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both involve save_pretrained behavior, but one is tokenizer config naming and the other is processor file emission; different code paths and fixes." + }, + { + "left": "issue:39692", + "right": "issue:46726", + "accept": false, + "reason": "SigLIP2 documentation example errors and AutoProcessor backend-missing errors are different problems, one in docs/example usage and one in runtime backend detection." + }, + { + "left": "issue:43065", + "right": "issue:46123", + "accept": false, + "reason": "SAM3 pixel decoder dummy layer behavior is unrelated to MaskGenerationPipeline dropping the last partial batch." + }, + { + "left": "issue:43475", + "right": "issue:46821", + "accept": false, + "reason": "A missing Sam3VisionEncoderOutput attribute and a TimesFM window_size AttributeError are unrelated model-specific bugs." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading failure and DeepSpeed ZeRO3 deque pop are distinct failures in different subsystems." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "Missing Sam3VisionEncoderOutput fields and AutoConfig trust_remote_code missing fields are not the same underlying bug." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving failure and Qwen2.5-GGUF loading failure are separate quantization/load-path issues." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Both are broad breakages, but one is custom model initialization on v5 and the other is an ESM2-specific regression; not the same fix." + }, + { + "left": "issue:46463", + "right": "issue:46877", + "accept": false, + "reason": "Incorrect T5 decoder validation messaging and auto_docstring stdout diagnostics are unrelated; one is model validation, the other is tooling noise." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "A placeholder bug report is not the same as a Gemma-4 ZeRO3 from_pretrained regression." + }, + { + "left": "issue:44373", + "right": "issue:46851", + "accept": false, + "reason": "Wrong docstring text for position_ids is documentation-only, while Qwen3.5 variable-length input failure is a runtime logic bug." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor save_pretrained file emission and deepspeed optimizer double-step behavior are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "PyTorch missing import error and Sentence Transformers loading hang have different symptoms and likely different root causes." + }, + { + "left": "issue:44871", + "right": "issue:46132", + "accept": false, + "reason": "Gemma eos_token_id config mismatch and AttentionInterface.register behavior changes are not the same code-path problem." + }, + { + "left": "issue:43723", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer loading in v5 and continue_final_message with mistral_common tokenizers are separate tokenizer issues." + }, + { + "left": "issue:44188", + "right": "issue:46851", + "accept": false, + "reason": "torch.compile branching divergence in attention kernels is unrelated to Qwen3.5 variable-length position_ids handling." + }, + { + "left": "issue:46877", + "right": "issue:47000", + "accept": false, + "reason": "Import-time doc-lint stdout noise is unrelated to RTDetrLoss infeasible cost-matrix crashes." + }, + { + "left": "issue:45561", + "right": "issue:47217", + "accept": false, + "reason": "pytest-xdist captured_info.txt race and VL image-placeholder handling are completely different bugs." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Generation kwargs update after sync_gpus and missing GenerationMixin cache in chunked_prefill are different generation-path regressions." + }, + { + "left": "issue:44451", + "right": "issue:47429", + "accept": false, + "reason": "Loading a specific model and an upper-bound packaging constraint on tokenizers are not the same underlying issue." + }, + { + "left": "issue:44279", + "right": "issue:47429", + "accept": false, + "reason": "A generic dependency issue and the tokenizers version bound problem are too different to merge as one fix." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe/v5 incompatibility and processor save_pretrained missing files are unrelated." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper word timestamp decoding crash and Kimi-K2.5 tokenizer regression affect different tokenization/decoding paths." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "GenerationMixin import failure and AutoImageProcessor import failure are both import errors, but for different symbols and subsystems." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Missing deepstack_visual_indexes in a config class and a Qwen3.5 Moe from_pretrained error are related by model family only, not by a shared defect." + }, + { + "left": "issue:44661", + "right": "issue:47059", + "accept": false, + "reason": "add-new-model-like tokenizer mapping failure and huggingface_hub CLI completion breakage are unrelated integration bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 51, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:29925", + "issue:30333", + "issue:30990", + "issue:33290", + "issue:35141", + "issue:41628", + "issue:42175", + "issue:42371", + "issue:42831", + "issue:42898", + "issue:43232", + "issue:43278", + "issue:43479", + "issue:43784", + "issue:43824", + "issue:43976", + "issue:44038", + "issue:44186", + "issue:44222", + "issue:44351", + "issue:44496", + "issue:44568", + "issue:44743", + "issue:44792", + "issue:44843", + "issue:44857", + "issue:44933", + "issue:44945", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45137", + "issue:45310", + "issue:45375", + "issue:45405", + "issue:45406", + "issue:45561", + "issue:45593", + "issue:45712", + "issue:45753", + "issue:45812", + "issue:46202", + "issue:46701", + "issue:46829", + "issue:46877", + "issue:47059", + "issue:47068", + "issue:47085" + ], + "soft_pairs": [ + "issue:41628|issue:44933", + "issue:43278|issue:45137", + "issue:45310|issue:45375", + "issue:45081|issue:45103", + "issue:29925|issue:30333", + "issue:42371|issue:45092", + "issue:44857|issue:46877", + "issue:45406|issue:46202", + "issue:46202|issue:46701", + "issue:45593|issue:46877", + "issue:43232|issue:44792", + "issue:43784|issue:43824", + "issue:42175|issue:44933", + "issue:44038|issue:45310", + "issue:41628|issue:44351", + "issue:44222|issue:45310", + "issue:30333|issue:33290", + "issue:45405|issue:46877", + "issue:44743|issue:47068", + "issue:42175|issue:45712", + "issue:42898|issue:44568", + "issue:44945|issue:45753", + "issue:45561|issue:47059", + "issue:42831|issue:44222", + "issue:42898|issue:45812", + "issue:44496|issue:45561", + "issue:30990|issue:35141", + "issue:43976|issue:44933", + "issue:44843|issue:47085", + "issue:44568|issue:47085", + "issue:43479|issue:46829", + "issue:44186|issue:47085" + ], + "prepared_review_unit_hash": "b4c3822ff0dc5b14464dcf8bc8907ecf703bd0339cd2f1bf41288be1a3b73664", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37373, + "estimated_input_tokens": 9344, + "estimated_eval_tokens": 18944 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This looks like a mostly heterogeneous set of Transformers issues with only superficial lexical similarity; I did not find any soft-edge pair that clearly represents the same underlying bug/change, so no duplicates were accepted.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44843 is the strongest standalone bug report here: it has a specific failing code path, clear offline-mode impact, and concrete reproducer details. It is still not a good canonical for the whole cluster because the rest are largely unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Different import failures in different modules; no shared code-path or concrete fix target." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Unrelated bugs: embedding dtype regression vs DeepSpeed ZeRO3 deque error." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Same model family, but one is a generic from_pretrained failure and the other is a specific config-field bug; not clearly the same defect." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Both are crashes, but in different utilities and different failure modes; not the same underlying bug." + }, + { + "left": "issue:29925", + "right": "issue:30333", + "accept": false, + "reason": "Completely unrelated: tokenizer fast-mode enforcement vs MLFlow reporting after failed training." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 configuration issue and meta-init incompatibility are different problems." + }, + { + "left": "issue:44857", + "right": "issue:46877", + "accept": false, + "reason": "Training-loss crash and docstring import-time logging are unrelated." + }, + { + "left": "issue:45406", + "right": "issue:46202", + "accept": false, + "reason": "Both involve Gemma, but one is a processor attribute bug and the other is a tokenizer roundtrip issue." + }, + { + "left": "issue:46202", + "right": "issue:46701", + "accept": false, + "reason": "Different pipeline failures with different symptoms and code paths." + }, + { + "left": "issue:45593", + "right": "issue:46877", + "accept": false, + "reason": "D-FINE loss behavior vs docstring diagnostics; no overlap in underlying bug." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation kwargs sync bug and Janus image-generation test failure are distinct." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different symbols and unrelated causes." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Packaging/backend dependency issue vs missing image_utils import; unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Same broad Qwen area, but different model variants and failure reports; not enough to treat as one concrete bug." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Different missing exports from transformers; no shared defect shown." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "FP8 save_pretrained behavior and Qwen3.5 loading error are different bugs." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "MLFlow reporting vs deepspeed OOM are unrelated." + }, + { + "left": "issue:45405", + "right": "issue:46877", + "accept": false, + "reason": "Dependency version pinning vs docstring import logging are unrelated." + }, + { + "left": "issue:44743", + "right": "issue:47068", + "accept": false, + "reason": "Recurrent-state reset bug and CUDA host-sync conditional are different generation issues." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "Dependency packaging issue vs leftover dummy classes leaking into dir(); unrelated." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both tokenizer regressions in v5, but they affect different APIs and different models." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel output corruption and a specific image infinite loop are not the same bug." + }, + { + "left": "issue:45561", + "right": "issue:47059", + "accept": false, + "reason": "Test xdist file race and CLI completion attribute error are unrelated test/runtime issues." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "Both mention FP8, but one is accuracy and the other is save_pretrained behavior; different issues." + }, + { + "left": "issue:42898", + "right": "issue:45812", + "accept": false, + "reason": "Different tokenizer regressions: cleanup-space behavior vs wrong token IDs." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Model loading/config error vs xdist race in testing utils; unrelated." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Loading stalls and embedding reinitialization are different model/setup bugs." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python-version compatibility issue vs missing import symbol; unrelated." + }, + { + "left": "issue:44843", + "right": "issue:47085", + "accept": false, + "reason": "Offline-mode Mistral regex patch bug and tokenizer thread-safety regression are different code paths." + }, + { + "left": "issue:44568", + "right": "issue:47085", + "accept": false, + "reason": "Tokenizer special-token regression and tokenizer thread-safety regression are not the same defect." + }, + { + "left": "issue:43479", + "right": "issue:46829", + "accept": false, + "reason": "Multimodal config defaulting bug vs incorrect hidden_states output in classification are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:47085", + "accept": false, + "reason": "LayoutLMv2Tokenizer crash and fast-tokenizer thread-safety regression are different bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 52, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:30333", + "issue:34689", + "issue:36331", + "issue:39692", + "issue:42222", + "issue:42673", + "issue:42796", + "issue:42898", + "issue:43116", + "issue:43295", + "issue:43479", + "issue:43704", + "issue:43723", + "issue:43874", + "issue:44016", + "issue:44060", + "issue:44079", + "issue:44190", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44448", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44869", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45200", + "issue:45310", + "issue:45405", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:46077", + "issue:46123", + "issue:46132", + "issue:46209", + "issue:46231", + "issue:46431", + "issue:46693", + "issue:46909", + "issue:47059", + "issue:47085" + ], + "soft_pairs": [ + "issue:44060|issue:45125", + "issue:42673|issue:46693", + "issue:42673|issue:43704", + "issue:44623|issue:45310", + "issue:39692|issue:46909", + "issue:43723|issue:44568", + "issue:30333|issue:34689", + "issue:30333|issue:36331", + "issue:43479|issue:44079", + "issue:45600|issue:47059", + "issue:42898|issue:43295", + "issue:42898|issue:44448", + "issue:42673|issue:47085", + "issue:42898|issue:44779", + "issue:43479|issue:44869", + "issue:44016|issue:46231", + "issue:22355|issue:28282", + "issue:46077|issue:46123", + "issue:42222|issue:45200", + "issue:46209|issue:46431", + "issue:44295|issue:45656", + "issue:44734|issue:45103", + "issue:42796|issue:43723", + "issue:43874|issue:45092", + "issue:45405|issue:45712", + "issue:43116|issue:44190", + "issue:44448|issue:46132", + "issue:44222|issue:44297" + ], + "prepared_review_unit_hash": "94dc4ae68a79adf7e6d1d0a267aae48681bd3e6dda1f0d9df38c119d987b34d2", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36284, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster contains one clear duplicate issue pair: issue:42673 and issue:43704 have the exact same title and describe the same Qwen3ForCausalLM VRAM leak when used in multiple dataloader threads. The rest of the soft pairs are thematic overlaps rather than the same underlying bug.", + "confidence": 0.97, + "canonical_issue_reason": "issue:42673 is the best canonical issue because it is the earlier report of the exact same VRAM-leak bug described again in issue:43704.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:42673 is the strongest representative issue in the cluster because it matches issue:43704 exactly in title and symptom, and it was created earlier.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Both are Qwen-family model issues, but they describe different failures: tied-weight warnings vs a missing tensor-parallel plan." + }, + { + "left": "issue:42673", + "right": "issue:46693", + "accept": false, + "reason": "Both concern resource/performance behavior, but one is a VRAM leak in dataloader threads and the other is a long-context FA2 regression. Different bugs." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Exact same title and same symptom: Qwen3ForCausalLM leaks VRAM when used in multiple dataloader threads. Clear duplicate." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "One is about processor.save_pretrained missing files; the other is a from_pretrained error for Qwen3.5 MoE. Related area, not the same bug." + }, + { + "left": "issue:39692", + "right": "issue:46909", + "accept": false, + "reason": "Both are documentation/example failures, but they involve different examples and different errors." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer regressions, but one is an AutoTokenizer loading issue in v5 and the other is missing BOS/EOS addition for a specific tokenizer." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Different subsystems and failure modes: MLflow reporting for failed training vs a model-loading regression for Llama 3.2 Vision." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "Unrelated issues: MLflow job reporting vs a CustomTrainer.compute_loss signature change." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Different bugs: multimodal config default initialization vs ModelOutput key assignment when a field was previously None." + }, + { + "left": "issue:45600", + "right": "issue:47059", + "accept": false, + "reason": "Different areas: stale auto_mappings config references vs a CLI failure with huggingface_hub 1.22.0." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Both are v5 regressions, but one is about clean_up_tokenization_spaces behavior and the other is processor.tokenizer access/images-to-tokenizer behavior." + }, + { + "left": "issue:42898", + "right": "issue:44448", + "accept": false, + "reason": "One is a tokenizer cleanup regression, the other is a Pegasus v4/v5 output difference. Not the same bug." + }, + { + "left": "issue:42673", + "right": "issue:47085", + "accept": false, + "reason": "Both involve threading/memory concerns, but one is a Qwen VRAM leak and the other is tokenizer thread-safety. Different underlying problem." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer-related, but Deepseek incorrect results and clean_up_tokenization_spaces behavior changes are distinct failures." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Completely different features: multimodal config defaults vs Whisper word-timestamp decoding crash." + }, + { + "left": "issue:44016", + "right": "issue:46231", + "accept": false, + "reason": "Different documentation problems: a notebook syntax error vs a parameter-name mismatch in set_training docs." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Both are import-related, but one is missing transformers.onnx and the other is missing PyTorch for AutoModel." + }, + { + "left": "issue:46077", + "right": "issue:46123", + "accept": false, + "reason": "Different code paths: Gemma4 attention-mask construction vs MaskGenerationPipeline final-batch handling." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "Different multimodal model bugs: broken vitpose models vs Gemma 4 mm_token_type_ids defaulting." + }, + { + "left": "issue:46209", + "right": "issue:46431", + "accept": false, + "reason": "Different runtime failures: FP8 inference on B200 vs apply_chat_template on a processor without a chat template." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Different failures: reading a position_ids buffer vs optimizer.step being called twice with deepspeed." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "One is a transformers serve KV-cache indexing crash; the other is an auto_docstring AttributeError with future annotations." + }, + { + "left": "issue:42796", + "right": "issue:43723", + "accept": false, + "reason": "Both are tokenizer-related, but one is BPE-marker handling and the other is AutoTokenizer.from_pretrained loading in v5." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Different model-specific issues: missing image-patches method in GLM46V processor vs remote-code incompatibility for old InternVL2 checkpoints." + }, + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "Unrelated maintenance issues: PEFT version pin bump vs leftover dummy classes leaking into repo checks." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Different example-script failures: empty multi-label classification results vs inability to load a local dataset." + }, + { + "left": "issue:44448", + "right": "issue:46132", + "accept": false, + "reason": "Different behavioral changes: Pegasus v4/v5 output mismatch vs AttentionInterface.register altering registered function behavior." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Different save_pretrained bugs: FP8 MoE save issues vs tokenizer_class mismatch in tokenizer_config.json." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The duplicate-pair conclusion for issue:42673 and issue:43704 is grounded, but the canonical/best-issue rationale overstates the evidence. The packet supports: exact same title, same symptom, and earlier creation time for issue:42673. It does not support \"more discussion/inbound references\" because inbound references are equal (0 vs 0), and \"more complete report\" is not directly evidenced in the packet. The soft-edge verdicts themselves are mostly conservative, but the summary should avoid those unsupported claims." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 53, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:30990", + "issue:38175", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42907", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43475", + "issue:43479", + "issue:43504", + "issue:43723", + "issue:44075", + "issue:44190", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44327", + "issue:44373", + "issue:44484", + "issue:44485", + "issue:44623", + "issue:44933", + "issue:44977", + "issue:45071", + "issue:45092", + "issue:45245", + "issue:45310", + "issue:45357", + "issue:45656", + "issue:45712", + "issue:45834", + "issue:46050", + "issue:46082", + "issue:46132", + "issue:46209", + "issue:46231", + "issue:46305", + "issue:46369", + "issue:46402", + "issue:46424", + "issue:46614", + "issue:46665", + "issue:46909", + "issue:46932", + "issue:47059", + "issue:47276" + ], + "soft_pairs": [ + "issue:43479|issue:46665", + "issue:44297|issue:45656", + "issue:42831|issue:44623", + "issue:44075|issue:46231", + "issue:44933|issue:45712", + "issue:45245|issue:46132", + "issue:44222|issue:44623", + "issue:44373|issue:44485", + "issue:43723|issue:44977", + "issue:43232|issue:47276", + "issue:44297|issue:45310", + "issue:43504|issue:47059", + "issue:44327|issue:47059", + "issue:46614|issue:46932", + "issue:43122|issue:46209", + "issue:43475|issue:46305", + "issue:41628|issue:42757", + "issue:43232|issue:45834", + "issue:30990|issue:38175", + "issue:45071|issue:46082", + "issue:45092|issue:45357", + "issue:44295|issue:44623", + "issue:44190|issue:46082", + "issue:44222|issue:44484", + "issue:46369|issue:46909", + "issue:43232|issue:46424", + "issue:42907|issue:46050", + "issue:43408|issue:46402" + ], + "prepared_review_unit_hash": "886ed0b27c8cf8c3a741a04261c0a25107a00c45ab1ca9c8f8c4671fb47e089b", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 35733, + "estimated_input_tokens": 8934, + "estimated_eval_tokens": 18124 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated issues spanning tokenization, config validation, serialization, FP8, SAM3, and dependency-compatibility paths. None of the soft pairs look like true duplicates of the same underlying bug or change.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue: the cluster is heterogeneous and the soft links do not indicate a shared underlying bug.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44297 is the clearest representative bug report: it is concrete, reproducible, and about a specific save_pretrained tokenizer metadata mismatch.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43479", + "right": "issue:46665", + "accept": false, + "reason": "Different problems: Phi4MultimodalConfig default initialization vs from_pretrained behavior inside Python classes." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save metadata mismatch and optimizer stepping twice in deepspeed are unrelated code paths." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy regression and missing processor save files are different serialization/compute issues." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "SGD args being ignored is a runtime bug; the other is only a doc parameter-name mismatch." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Both mention imports, but one is a missing image_utils export and the other is cleanup of dummy classes." + }, + { + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Category-count limit error and AttentionInterface.register behavior are unrelated." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both involve save_pretrained, but one is FP8 MOE saving and the other is processor file omission." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Docstring typo for position_ids is unrelated to max_shard_size behavior in save_pretrained." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading in v5 and Qwen3.5 flash-attention generation are different failure modes." + }, + { + "left": "issue:43232", + "right": "issue:47276", + "accept": false, + "reason": "Both mention synced_gpus, but one is generation state update logic and the other is a warning message." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer_class mismatch and Qwen3.5 MoE from_pretrained error are different bugs." + }, + { + "left": "issue:43504", + "right": "issue:47059", + "accept": false, + "reason": "A pretrained model legacy-field loading bug is unrelated to a huggingface_hub CLI incompatibility." + }, + { + "left": "issue:44327", + "right": "issue:47059", + "accept": false, + "reason": "QA span decoding kth-out-of-bounds and CLI _add_completion missing are unrelated." + }, + { + "left": "issue:46614", + "right": "issue:46932", + "accept": false, + "reason": "Stale phonemizer backend state and Gemma4 quantized embedder dtype casting are different subsystems." + }, + { + "left": "issue:43122", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer regression across versions is unrelated to FP8 inference failure on B200." + }, + { + "left": "issue:43475", + "right": "issue:46305", + "accept": false, + "reason": "Missing SAM3 vision encoder attribute and MoonshineStreaming off-by-one are different code paths." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Two separate import errors for different symbols and dependencies; not the same underlying fix." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Generation sync_gpus state handling and Kosmos2.5 OCR index error are unrelated." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Sentence Transformers loading hang and siglip2 zero probabilities are different model-specific issues." + }, + { + "left": "issue:45071", + "right": "issue:46082", + "accept": false, + "reason": "PretrainedConfig type checking break and LlamaConfig head_dim validation are distinct config bugs." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Remote-code meta-init incompatibility and Qwen3.5 visual-key saving regression are not the same bug." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Buffer access error for position_ids and processor.save_pretrained missing files are unrelated." + }, + { + "left": "issue:44190", + "right": "issue:46082", + "accept": false, + "reason": "Local dataset loading in an example script is unrelated to LlamaConfig head_dim validation." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "FP8 MOE save bug and the 50GB max_shard_size question are not the same issue." + }, + { + "left": "issue:46369", + "right": "issue:46909", + "accept": false, + "reason": "SAM3 memory overflow and a broken DIA documentation example are unrelated." + }, + { + "left": "issue:45071", + "right": "issue:45092", + "accept": false, + "reason": "Both are v5 migration issues, but one is type-checking and the other is remote-code checkpoint meta initialization." + }, + { + "left": "issue:44295", + "right": "issue:46424", + "accept": false, + "reason": "position_ids buffer read failure and cache max_cache_length semantics are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:46050", + "accept": false, + "reason": "Dequantized model save failure and sinq save/load breakage are both serialization-related, but for different quantization paths and symptoms." + }, + { + "left": "issue:43408", + "right": "issue:46402", + "accept": false, + "reason": "SAM3 tracker warning and SAM3 text-encoder weights not loading are different model-loading problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 54, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:28218", + "issue:39379", + "issue:39401", + "issue:41628", + "issue:42796", + "issue:42831", + "issue:42886", + "issue:43066", + "issue:43122", + "issue:43408", + "issue:43540", + "issue:43576", + "issue:43644", + "issue:43756", + "issue:43784", + "issue:43824", + "issue:43986", + "issue:44038", + "issue:44060", + "issue:44188", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44360", + "issue:44373", + "issue:44483", + "issue:44568", + "issue:44829", + "issue:44964", + "issue:45092", + "issue:45310", + "issue:45381", + "issue:45406", + "issue:45600", + "issue:45753", + "issue:46002", + "issue:46172", + "issue:46305", + "issue:46328", + "issue:46369", + "issue:46655", + "issue:46726", + "issue:46762", + "issue:46909", + "issue:47059", + "issue:47217" + ], + "soft_pairs": [ + "issue:43986|issue:46726", + "issue:44360|issue:46762", + "issue:44263|issue:45092", + "issue:45381|issue:46305", + "issue:44263|issue:44373", + "issue:44279|issue:46328", + "issue:22355|issue:27957", + "issue:41628|issue:43824", + "issue:46305|issue:46655", + "issue:43784|issue:45600", + "issue:22355|issue:28218", + "issue:42831|issue:45310", + "issue:44483|issue:45406", + "issue:43576|issue:43784", + "issue:44964|issue:47059", + "issue:43756|issue:44060", + "issue:42831|issue:44038", + "issue:44188|issue:44373", + "issue:44261|issue:45092", + "issue:44263|issue:44360", + "issue:44829|issue:46002", + "issue:43986|issue:47217", + "issue:43644|issue:46305", + "issue:42796|issue:43122", + "issue:42831|issue:45753", + "issue:43540|issue:46909", + "issue:39401|issue:44568", + "issue:43408|issue:46369", + "issue:45381|issue:46909", + "issue:42886|issue:43066", + "issue:39379|issue:46172" + ], + "prepared_review_unit_hash": "249533a50a3de758b398d23cb098a37545b262c1f7dd343012b3c00c77dd373d", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36752, + "estimated_input_tokens": 9188, + "estimated_eval_tokens": 18632 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: most items are only loosely similar tokenizer/vision/model regression reports, with no clear same-bug duplicate pair. The closest matches still differ in concrete code path or model/component, so I would not merge any of the soft edges.", + "confidence": 0.29, + "canonical_issue_reason": "Issue 43122 is the broadest version-level tokenizer regression in the set and the best representative of the tokenizer subtheme, though the cluster overall is too mixed for a strong canonical choice.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43122 is the most general and reusable issue title among the tokenizer-related reports; it best captures a family of tokenization regressions, even if this cluster is not a clean duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43986", + "right": "issue:46726", + "accept": false, + "reason": "Both are AutoProcessor/vision-backend failures, but they describe different missing-backend behaviors and different error paths." + }, + { + "left": "issue:44360", + "right": "issue:46762", + "accept": false, + "reason": "Different bugs: one is a missing ReLU in the DSA indexer, the other is numerical divergence in MSA attention." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "Unrelated code paths: torch.split behavior in GlmMoeDsaIndexer vs remote-code/meta-init incompatibility in InternVL2." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Both involve video processing, but one is wrong vision_position_ids and the other is a sliding-window mask off-by-one." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "A runtime tensor-splitting bug and a docstring issue are not the same underlying bug." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "Generic dependency issue vs a padding performance inefficiency; not the same defect." + }, + { + "left": "issue:22355", + "right": "issue:27957", + "accept": false, + "reason": "Different import/model issues: missing transformers.onnx module vs XLMRoberta Flash Attention 2 support." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different missing exports/code paths." + }, + { + "left": "issue:46305", + "right": "issue:46655", + "accept": false, + "reason": "Both are video-related, but one is a mask off-by-one and the other is a padding-calculation error." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Different failures: missing nn import in sentence-transformers integration vs stale auto_mappings references to removed configs." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "Different tokenizer/module problems; one is a missing module import, the other is an extra-space tokenization bug." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "Both mention model/precision regressions, but the concrete failures and affected components differ." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Chat/completions request handling is unrelated to Gemma4Processor's missing _tokenizer crash." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "CLI env command breakage and an import-time NameError are distinct failures." + }, + { + "left": "issue:44964", + "right": "issue:47059", + "accept": false, + "reason": "Model loading failure and CLI test failure with huggingface_hub are different issues." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch and Qwen3-Next tied-weights warning are separate model-specific bugs." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression is not the same as the Qwen3-VL-Moe bug." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "Attention-kernel divergence and a wrong docstring are not the same bug." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "Different problems: missing rms_norm_eps in MLA q_a_layernorm vs meta-init incompatibility for old InternVL2 checkpoints." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Same model family, but different functions and symptoms; not one underlying defect." + }, + { + "left": "issue:44829", + "right": "issue:46002", + "accept": false, + "reason": "Training instability under flash_attention_3 and NaNs after checkout/init are not the same concrete bug." + }, + { + "left": "issue:43986", + "right": "issue:47217", + "accept": false, + "reason": "Both are VL/processor issues, but one is about missing torchvision during loading and the other about image placeholders in user text." + }, + { + "left": "issue:43644", + "right": "issue:46305", + "accept": false, + "reason": "Non-persistent buffer initialization junk and a video mask off-by-one are unrelated." + }, + { + "left": "issue:42796", + "right": "issue:43122", + "accept": false, + "reason": "Tokenizer BPE-marker handling and broader v4-to-v5 tokenization differences are related only at a high level, not the same bug." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "Precision/accuracy issue and an infinite-loop image-triggered bug are different failure modes." + }, + { + "left": "issue:43540", + "right": "issue:46909", + "accept": false, + "reason": "Qwen3OmniMoe video-input processing failure is unrelated to a broken documentation example." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Wrong offset_mapping and missing BOS/EOS insertion are different tokenizer defects." + }, + { + "left": "issue:43408", + "right": "issue:46369", + "accept": false, + "reason": "Model-type warning on instantiation and memory overflow in sam3-litetext are not the same issue." + }, + { + "left": "issue:45381", + "right": "issue:46909", + "accept": false, + "reason": "A multimodal video-position bug and a docs example bug do not share the same underlying fix." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache loading and wrong tokenizer decoder type are separate tokenizer-related regressions." + }, + { + "left": "issue:39379", + "right": "issue:46172", + "accept": false, + "reason": "Different tensor/masking bugs in different models; too far apart to be duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 55, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:29159", + "issue:30990", + "issue:33357", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42757", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43116", + "issue:43122", + "issue:43329", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43646", + "issue:43824", + "issue:43906", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44117", + "issue:44222", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44488", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44829", + "issue:44945", + "issue:44964", + "issue:44995", + "issue:45200", + "issue:45216", + "issue:45278", + "issue:45290", + "issue:45478", + "issue:46002", + "issue:46326", + "issue:46697", + "issue:47059" + ], + "soft_pairs": [ + "issue:42886|issue:42994", + "issue:27957|issue:30990", + "issue:42757|issue:45278", + "issue:43646|issue:46002", + "issue:30990|issue:33357", + "issue:43329|issue:47059", + "issue:43122|issue:46326", + "issue:44625|issue:44829", + "issue:44295|issue:44297", + "issue:22355|issue:29159", + "issue:42907|issue:44964", + "issue:44117|issue:44704", + "issue:39401|issue:44521", + "issue:41628|issue:45278", + "issue:39401|issue:44488", + "issue:42907|issue:44945", + "issue:43976|issue:44945", + "issue:43122|issue:44521", + "issue:44704|issue:45290", + "issue:42907|issue:43452", + "issue:43408|issue:43475", + "issue:42371|issue:44263", + "issue:43824|issue:44351", + "issue:42907|issue:43937", + "issue:44038|issue:44297", + "issue:43116|issue:45200", + "issue:45216|issue:45478", + "issue:44038|issue:44295", + "issue:44038|issue:44222", + "issue:44995|issue:46697", + "issue:43122|issue:43906" + ], + "prepared_review_unit_hash": "3cd174b84293227eed160d6229b2b0da048469d4cc85d71a7b1d0d23dd627e0e", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36036, + "estimated_input_tokens": 9009, + "estimated_eval_tokens": 18274 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most candidates are clearly unrelated issues spanning different models, features, and failure modes. The only plausible duplicate-like pair is the Qwen3.5 save/load regression around transformers 5.4.0, which appears to describe the same underlying serialization bug from two directions.", + "confidence": 0.89, + "canonical_issue_reason": "issue:45216 is the strongest representative issue here: it describes a concrete, version-bounded Qwen3.5 regression in save/load behavior and is the clearest root-cause-style report among the only plausible duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45216 is the best standalone issue candidate because it is specific, reproducible, and tied to a clear regression boundary (since 5.4.0).", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Both involve saving/loading, but one is tokenizer cache loading offline and the other is quantized model saving; different code paths and failure modes." + }, + { + "left": "issue:27957", + "right": "issue:30990", + "accept": false, + "reason": "Different problems: Flash Attention 2 support for XLM-Roberta vs Sentence Transformers getting stuck while loading." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "The second is a broad post-upgrade import-error report, not the same concrete missing symbol bug as is_offline_mode." + }, + { + "left": "issue:43646", + "right": "issue:46002", + "accept": false, + "reason": "Custom model initialization breakage and NaNs in classification heads are distinct bugs with different symptoms." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Loading hang for Sentence Transformers and a MacOS bus error on a CLIP model are unrelated failures." + }, + { + "left": "issue:43329", + "right": "issue:47059", + "accept": false, + "reason": "Multimodal token counting bug vs CLI incompatibility with huggingface_hub; different subsystems entirely." + }, + { + "left": "issue:43122", + "right": "issue:46326", + "accept": false, + "reason": "Both are tokenizer-related, but one is an offset/differences regression and the other is a mistral_common chat-template bug." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and flash_attention_3 training degeneration are unrelated issues." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "One is a position_ids buffer/read problem, the other is tokenizer_class metadata written by save_pretrained." + }, + { + "left": "issue:22355", + "right": "issue:29159", + "accept": false, + "reason": "Missing transformers.onnx import and slow/fast tokenizer inconsistency are different topics." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Quantized/dequantized model saving and loading a multimodal Phi model are unrelated." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES returning None and AutoProcessor kwargs not reaching cached_file affect different layers of the loading stack." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Wrong offset_mapping and zero assistant_masks in multimodal chat templates are different tokenizer behaviors." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "The latter is a broad upgrade regression, not the same specific AutoImageProcessor import failure." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Offset mapping bugs and failure to load a specific tokenizer are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Quantized model save failure and incorrect LLM output under pipeline parallelism do not share the same code path." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version compatibility and pipeline-parallel output corruption are different bugs." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "General tokenization differences versus multimodal assistant mask generation are not the same underlying bug." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "AutoProcessor cached_file forwarding and apply_chat_template crashing on tool-call messages are distinct issues." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Dequantized model saving and gguf_file loading regressions are separate features and failure modes." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Both mention SAM 3 video, but one is a model-type warning and the other is a missing attribute during vision encoder output handling; not the same bug." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 settings warning and GlmMoeDsa torch.split behavior are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports: Qwen2_5_VLForConditionalGeneration vs HybridCache." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Quantized save failure and invalid GenerationConfig for GLM-5 are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe compatibility and tokenizer_class metadata mismatch are separate problems." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Example-script multi-label classification behavior and Gemma 4 mm_token_type_ids defaults are different concerns." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": true, + "reason": "Both are Qwen3.5 regressions tied to transformers 5.4.0+ save/load behavior; one reports an incorrect saved checkpoint and the other a from_pretrained failure, which plausibly share the same underlying serialization/config bug." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Qwen3-VL-Moe version breakage and a position_ids buffer access issue are unrelated." + }, + { + "left": "issue:44995", + "right": "issue:46697", + "accept": false, + "reason": "GlmMoeDsa stale cache on second forward pass and Swin2SR over-padding are different model-specific bugs." + }, + { + "left": "issue:43122", + "right": "issue:43906", + "accept": false, + "reason": "A tokenization regression and a reproduction of an unrelated external issue are not the same bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: issue:45216 is indeed the clearest version-bounded Qwen3.5 save/load regression candidate by title, and the only accepted soft-edge pair (45216/45478) is a conservative same-area match rather than an overreach. The remaining verdicts are appropriately cautious and the reasons stay within the evidence provided by the titles." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 56, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:22355", + "issue:28348", + "issue:29127", + "issue:36296", + "issue:38617", + "issue:39379", + "issue:39401", + "issue:42175", + "issue:42222", + "issue:42673", + "issue:42757", + "issue:42898", + "issue:43066", + "issue:43116", + "issue:43232", + "issue:43278", + "issue:43493", + "issue:43611", + "issue:43976", + "issue:44188", + "issue:44351", + "issue:44360", + "issue:44483", + "issue:44485", + "issue:44610", + "issue:44625", + "issue:44779", + "issue:44938", + "issue:44945", + "issue:45072", + "issue:45083", + "issue:45278", + "issue:45357", + "issue:45446", + "issue:45478", + "issue:45753", + "issue:46002", + "issue:46172", + "issue:46231", + "issue:46328", + "issue:46614", + "issue:46640", + "issue:46762", + "issue:46877", + "issue:46909", + "issue:47217", + "issue:47312" + ], + "soft_pairs": [ + "issue:44360|issue:44485", + "issue:43278|issue:44945", + "issue:22355|issue:29127", + "issue:46172|issue:46762", + "issue:42175|issue:44938", + "issue:28348|issue:46640", + "issue:42222|issue:47312", + "issue:39379|issue:43232", + "issue:43493|issue:45753", + "issue:39401|issue:46614", + "issue:42757|issue:44351", + "issue:44188|issue:46328", + "issue:42673|issue:45072", + "issue:36296|issue:43116", + "issue:45357|issue:45478", + "issue:46231|issue:46877", + "issue:44483|issue:47217", + "issue:42898|issue:46614", + "issue:43976|issue:45446", + "issue:44625|issue:46002", + "issue:44610|issue:46909", + "issue:45083|issue:45446", + "issue:43611|issue:46002", + "issue:38617|issue:45278", + "issue:43116|issue:43493", + "issue:43066|issue:44779" + ], + "prepared_review_unit_hash": "0f15dc0c3e0d87950a575ef5ed0f2a01babc582ba4215787f7ae839357985066", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 34379, + "estimated_input_tokens": 8595, + "estimated_eval_tokens": 17446 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is a heterogeneous mix of unrelated bugs, regressions, and docs/import issues. None of the soft pairs look like the same underlying change or bug, so I reject all soft edges. The broadest issue is the multi-import-regression report, which is the best umbrella representative here.", + "confidence": 0.92, + "canonical_issue_reason": "issue:45278 is the broadest, most umbrella-style report in the set ('many import errors after update'), making it the best canonical issue candidate among otherwise unrelated reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the most general and reusable problem statement in the cluster; it can serve as the closest thing to a representative issue even though the cluster is not a true duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "Unrelated topics: DSA indexer/ReLU discussion vs GLM-5 RoPE implementation." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Different bug classes: dtype drift between train/eval vs incorrect output under pipeline parallelism." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "Import failure for transformers.onnx is unrelated to LayoutLMv3 box-content error messaging." + }, + { + "left": "issue:46172", + "right": "issue:46762", + "accept": false, + "reason": "Different models and failure modes: Deepseek-v4 tensor expression vs Minimax attention numerical mismatch." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Both are environment/runtime problems, but one is missing TensorFlow from install extras and the other is a Python 3.14 load failure; not the same bug." + }, + { + "left": "issue:28348", + "right": "issue:46640", + "accept": false, + "reason": "Both mention flash attention 2, but one is a GPT-2 support request and the other is an unspecified issue; too vague to merge as the same bug." + }, + { + "left": "issue:42222", + "right": "issue:47312", + "accept": false, + "reason": "Vitpose model breakage is unrelated to kernel compatibility with 0.16." + }, + { + "left": "issue:39379", + "right": "issue:43232", + "accept": false, + "reason": "Mllama masking logic and generation kwargs after sync_gpus are separate code paths." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Different models and symptoms: SigLIP2 JAX parity mismatch vs Qwen3_5 image-triggered infinite loop." + }, + { + "left": "issue:39401", + "right": "issue:46614", + "accept": false, + "reason": "Tokenizer offset mapping bug is unrelated to stale espeak backend language state in Wav2Vec2PhonemeCTCTokenizer." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different missing symbols from different packages; not the same concrete failure." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "Attention-kernel divergence under torch.compile is unrelated to a padding performance issue." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM is unrelated to dtype mismatches in inference tests." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training bug and empty multi-label results in an example script are different behaviors with no concrete shared code path." + }, + { + "left": "issue:45357", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5, but one is save_pretrained key serialization and the other is from_pretrained loading failure; too different to be duplicates." + }, + { + "left": "issue:46231", + "right": "issue:46877", + "accept": false, + "reason": "Documentation parameter name mismatch is unrelated to auto_docstring printing diagnostics at import time." + }, + { + "left": "issue:44483", + "right": "issue:47217", + "accept": false, + "reason": "Chat completions request rejection is not the same as VL models failing on image placeholders in user text." + }, + { + "left": "issue:42898", + "right": "issue:46614", + "accept": false, + "reason": "Tokenization cleanup behavior in v5 is unrelated to stale phonemizer backend language switching." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Both are version-compatibility checks, but they concern different runtimes and different import sites." + }, + { + "left": "issue:44625", + "right": "issue:46002", + "accept": false, + "reason": "Configuration propagation bug in Qwen3.5 is unrelated to NaNs in classification heads after checkout/init." + }, + { + "left": "issue:44610", + "right": "issue:46909", + "accept": false, + "reason": "Processor input-size mismatch is unrelated to a broken documentation example." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "Helper output-length logic in qwen3_omni_moe is unrelated to a PyTorch version gate for flex_attention." + }, + { + "left": "issue:43611", + "right": "issue:46002", + "accept": false, + "reason": "Base-model-prefix loading regression and NaNs in classification heads are not the same failure." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "The latter is a broad umbrella of many import regressions; the former is one specific missing symbol and not clearly the same bug." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Example-script classification output issue is unrelated to SigLIP2 implementation parity." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both involve tokenizers and v5 regressions, but one is decoder-type selection and the other is Deepseek output correctness; not enough evidence they are the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 57, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:33453", + "issue:39379", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42617", + "issue:42673", + "issue:42796", + "issue:43064", + "issue:43066", + "issue:43116", + "issue:43208", + "issue:43618", + "issue:43673", + "issue:43688", + "issue:43824", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43937", + "issue:43992", + "issue:44075", + "issue:44117", + "issue:44291", + "issue:44355", + "issue:44485", + "issue:44704", + "issue:44936", + "issue:45071", + "issue:45081", + "issue:45092", + "issue:45341", + "issue:45362", + "issue:45405", + "issue:45446", + "issue:45478", + "issue:45563", + "issue:45588", + "issue:46231", + "issue:46328", + "issue:46710", + "issue:46726", + "issue:46762", + "issue:46829", + "issue:46894", + "issue:46981", + "issue:47000", + "issue:47116" + ], + "soft_pairs": [ + "issue:44075|issue:44936", + "issue:45362|issue:45478", + "issue:43937|issue:44485", + "issue:42673|issue:42796", + "issue:43066|issue:46710", + "issue:46726|issue:46981", + "issue:44355|issue:46231", + "issue:43618|issue:46829", + "issue:43116|issue:43825", + "issue:42371|issue:43874", + "issue:43827|issue:45446", + "issue:43992|issue:44704", + "issue:45405|issue:45446", + "issue:44117|issue:45588", + "issue:43824|issue:46981", + "issue:44291|issue:45071", + "issue:39379|issue:43688", + "issue:46328|issue:47116", + "issue:44117|issue:45081", + "issue:33453|issue:39401", + "issue:43673|issue:46328", + "issue:43208|issue:46894", + "issue:43064|issue:43673", + "issue:43937|issue:45563", + "issue:42617|issue:45341", + "issue:42222|issue:47000", + "issue:45092|issue:46762" + ], + "prepared_review_unit_hash": "fb6288bf0bcd4d785168287fa355dfc337abaccea3a49ff5955774daa22e0c89", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35627, + "estimated_input_tokens": 8907, + "estimated_eval_tokens": 18070 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most pairs are only loosely related by broad areas like tokenizers, v5 regressions, or model loading, but they do not describe the same underlying bug or change.", + "confidence": 0.27, + "canonical_issue_reason": "No single true canonical duplicate stands out; issue:44117 is only the best broad representative because it concerns a generic from_pretrained/tokenizer-loading assumption, while most others are model-specific or unrelated.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44117 is the most reusable cluster anchor: it targets a core tokenizer-loading path and is broader than the other mostly narrow, model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "Different failures in different training/eval paths: unused SGD args vs evaluate() breaking after train()." + }, + { + "left": "issue:45362", + "right": "issue:45478", + "accept": false, + "reason": "Both mention Qwen3.5, but one is a chat crash and the other is a from_pretrained MoE loading error; not the same bug." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "One is invalid GenerationConfig handling, the other is RoPE implementation discussion; separate GLM-5 issues." + }, + { + "left": "issue:42673", + "right": "issue:42796", + "accept": false, + "reason": "VRAM leak in multi-threaded dataloading vs tokenizer BPE marker handling; unrelated subsystems." + }, + { + "left": "issue:43066", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer decoder type regression and an accuracy regression in a specific model are not the same code-path problem." + }, + { + "left": "issue:46726", + "right": "issue:46981", + "accept": false, + "reason": "AutoProcessor missing-vision-backend error and repeated HTTP 429 retries are different loading/network behaviors." + }, + { + "left": "issue:44355", + "right": "issue:46231", + "accept": false, + "reason": "Compiled Python file errors and a set_training() docs parameter mismatch are unrelated." + }, + { + "left": "issue:43618", + "right": "issue:46829", + "accept": false, + "reason": "Both concern output fields, but they affect different model heads and different output objects; not one duplicate bug." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Example script multi-label empty results and a pipeline() error message about translation support are distinct issues." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 API-setting guidance and a missing image-patch method in GLM46V are unrelated." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs still referencing pipeline() and a PyTorch version-check bug in flex_attention are different kinds of defects." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Both involve from_pretrained, but one misses model weights and the other fails to forward kwargs to cached_file." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "A dependency version bump to an unreleased PEFT version is unrelated to a flex_attention import/version check." + }, + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "Tokenizer mapping None handling and a flash_attention crash on s_aux=None are separate bugs in different paths." + }, + { + "left": "issue:43824", + "right": "issue:46981", + "accept": false, + "reason": "Importing a missing model class and retry behavior on 429 responses are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "Both are v5 loading-related, but one is init_empty_weights/_is_hf_initialized and the other is PretrainedConfig type checking; not the same fix." + }, + { + "left": "issue:39379", + "right": "issue:43688", + "accept": false, + "reason": "mllama mask handling and MoE auxiliary-loss normalization are different model-specific math bugs." + }, + { + "left": "issue:46328", + "right": "issue:47116", + "accept": false, + "reason": "A SequenceFeatureExtractor.pad performance issue is unrelated to a streaming pipeline regression." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer mapping fallback and Mistral regex patch crashes involve different tokenizer code paths." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Both are tokenizer-related, but one is loading regression and the other is offset_mapping for Qwen3; not the same underlying defect." + }, + { + "left": "issue:43673", + "right": "issue:46328", + "accept": false, + "reason": "Generation cache handling and feature extractor padding performance are unrelated." + }, + { + "left": "issue:43208", + "right": "issue:46894", + "accept": false, + "reason": "xLSTM training blockers and a CohereASR double-shift loss bug are not the same issue, even if both affect training." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Optimizer-state corruption under FSDP2/PEFT is unrelated to missing GenerationMixin cache during chunked prefill." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Invalid generation config and a stale warning for num_return_sequences are separate generation issues." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "3d_parallel.py runtime failure and a small testing_utils bug do not share an underlying cause." + }, + { + "left": "issue:42222", + "right": "issue:47000", + "accept": false, + "reason": "VitPose model breakage and RTDetrLoss infeasible cost matrix errors are unrelated model failures." + }, + { + "left": "issue:45092", + "right": "issue:46762", + "accept": false, + "reason": "Meta-initialization incompatibility for old InternVL2 checkpoints and Minimax attention numerical drift are different problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 58, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:38617", + "issue:39379", + "issue:39401", + "issue:41628", + "issue:42222", + "issue:43097", + "issue:43122", + "issue:43408", + "issue:43421", + "issue:43441", + "issue:43450", + "issue:43742", + "issue:43937", + "issue:43986", + "issue:44220", + "issue:44263", + "issue:44380", + "issue:44410", + "issue:44466", + "issue:44483", + "issue:44493", + "issue:44496", + "issue:44704", + "issue:44779", + "issue:44829", + "issue:44877", + "issue:44964", + "issue:44977", + "issue:45071", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45464", + "issue:45561", + "issue:45632", + "issue:46002", + "issue:46077", + "issue:46231", + "issue:46298", + "issue:46328", + "issue:46512", + "issue:46752", + "issue:46762", + "issue:46897", + "issue:46981", + "issue:47059", + "issue:47110" + ], + "soft_pairs": [ + "issue:44466|issue:46002", + "issue:44380|issue:46077", + "issue:43408|issue:43742", + "issue:43937|issue:44977", + "issue:43450|issue:43986", + "issue:42222|issue:43441", + "issue:39379|issue:46897", + "issue:45071|issue:46512", + "issue:42222|issue:46752", + "issue:44263|issue:46762", + "issue:44877|issue:46231", + "issue:39401|issue:44779", + "issue:44829|issue:45216", + "issue:45632|issue:46981", + "issue:45341|issue:45561", + "issue:44410|issue:45310", + "issue:45071|issue:47059", + "issue:45341|issue:46328", + "issue:39401|issue:43421", + "issue:44483|issue:45464", + "issue:38617|issue:41628", + "issue:38617|issue:44220", + "issue:43937|issue:44496", + "issue:44704|issue:47110", + "issue:43122|issue:46298", + "issue:44964|issue:45278", + "issue:43097|issue:44493" + ], + "prepared_review_unit_hash": "e7c9e3d840222e644bab457bfe24b9a758e0bf0e0d2a172bf43e75548662f5aa", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35610, + "estimated_input_tokens": 8903, + "estimated_eval_tokens": 18062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is highly heterogeneous: it mixes unrelated tokenization, model-loading, config, processor, API, and training bugs. The soft links mostly reflect broad transformer-v5 regression wording or shared subsystem names, not the same underlying defect.", + "confidence": 0.96, + "canonical_issue_reason": "No single issue is a true duplicate of the whole cluster; if forced to pick an umbrella, issue:45278 is the broadest regression report, but it still only loosely covers a subset of the import/load failures.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best umbrella candidate because it is the most general user-facing regression report and could absorb some related import/loading complaints, even though most items here are distinct bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44466", + "right": "issue:46002", + "accept": false, + "reason": "Different symptoms and code paths: weight serialization/tied weights vs NaNs in classification heads." + }, + { + "left": "issue:44380", + "right": "issue:46077", + "accept": false, + "reason": "Unrelated model bugs: GPT-2 attention scaling vs Gemma4 bidirectional mask construction." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "A warning about model type mismatch is not the same bug as a key error while loading a MobileLLM model." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "Different failures: invalid GenerationConfig vs flash-attention generation behavior for Qwen3.5." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Batched video processor shape bug and missing-torchvision crash are separate problems." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Vitpose breakage and Ministral FlashAttention failure are unrelated model-specific regressions." + }, + { + "left": "issue:39379", + "right": "issue:46897", + "accept": false, + "reason": "mllama mask handling and Florence2 training-loss shifting are different bugs, only loosely pattern-matched." + }, + { + "left": "issue:45071", + "right": "issue:46512", + "accept": false, + "reason": "Type-checking regression in v5.4.0 is not the same as stale config.dtype after from_config(dtype=...)." + }, + { + "left": "issue:42222", + "right": "issue:46752", + "accept": false, + "reason": "Vitpose model breakage and empty chat-template handling are unrelated." + }, + { + "left": "issue:44263", + "right": "issue:46762", + "accept": false, + "reason": "torch.split return semantics in an indexer is not the same as numerical drift in Minimax attention." + }, + { + "left": "issue:44877", + "right": "issue:46231", + "accept": false, + "reason": "Strict config loading and a documentation parameter-name typo are not the same defect." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer-related, but they affect different models and different failure modes." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "FlashAttention training degeneration and save_pretrained checkpoint corruption are separate issues." + }, + { + "left": "issue:45632", + "right": "issue:46981", + "accept": false, + "reason": "Cache-path collision for local models is unrelated to repeated retries on HTTP 429 responses." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "A small testing_utils bug and an xdist race on captured_info.txt are not the same underlying issue." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Missing Qwen3Next projections and Qwen3.5 MoE from_pretrained failure are different model-loading problems." + }, + { + "left": "issue:45071", + "right": "issue:47059", + "accept": false, + "reason": "PretrainedConfig type checking and a CLI completion attribute error are unrelated." + }, + { + "left": "issue:45341", + "right": "issue:46328", + "accept": false, + "reason": "Testing-utils bug versus a pad performance inefficiency; same file area, different bug." + }, + { + "left": "issue:39401", + "right": "issue:43421", + "accept": false, + "reason": "Wrong offset_mapping and runtime post-processor updates are different tokenizer behaviors." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both mention chat/completions, but one is request acceptance and the other is streaming inference failure on a specific model." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Different missing imports: layer_type_validation vs AutoImageProcessor." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "ImportError for a config utility is unrelated to a feature-extraction function issue." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "Invalid GenerationConfig and missing model_type in config.json are distinct loading/configuration errors." + }, + { + "left": "issue:44704", + "right": "issue:47110", + "accept": false, + "reason": "Forwarding kwargs to cached_file and special_tokens list handling are not the same code-path problem." + }, + { + "left": "issue:43122", + "right": "issue:46298", + "accept": false, + "reason": "A tokenizer regression is not a noisy logging message bug." + }, + { + "left": "issue:44964", + "right": "issue:45278", + "accept": false, + "reason": "A multimodal model load failure and a broad import-error report overlap only at the surface level." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Removed config field and unexpected position-id key errors are different regression types." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 59, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:27957", + "issue:29925", + "issue:30947", + "issue:42371", + "issue:42831", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43122", + "issue:43208", + "issue:43352", + "issue:43441", + "issue:43452", + "issue:43531", + "issue:43577", + "issue:43582", + "issue:43673", + "issue:43874", + "issue:43931", + "issue:43957", + "issue:44038", + "issue:44075", + "issue:44077", + "issue:44117", + "issue:44230", + "issue:44261", + "issue:44263", + "issue:44351", + "issue:44360", + "issue:44410", + "issue:44484", + "issue:44485", + "issue:44829", + "issue:44987", + "issue:45071", + "issue:45310", + "issue:45341", + "issue:45478", + "issue:45542", + "issue:45588", + "issue:45656", + "issue:46050", + "issue:46172", + "issue:46752", + "issue:46762", + "issue:46897", + "issue:46981", + "issue:47116" + ], + "soft_pairs": [ + "issue:42371|issue:44261", + "issue:42371|issue:42831", + "issue:44261|issue:46762", + "issue:43208|issue:46897", + "issue:44077|issue:46981", + "issue:43931|issue:44410", + "issue:44484|issue:45310", + "issue:44075|issue:45656", + "issue:43066|issue:43122", + "issue:44410|issue:45478", + "issue:44263|issue:44485", + "issue:29925|issue:30947", + "issue:43673|issue:45341", + "issue:43097|issue:43452", + "issue:43931|issue:44038", + "issue:45071|issue:45542", + "issue:44038|issue:44230", + "issue:43441|issue:44829", + "issue:43012|issue:43577", + "issue:43957|issue:44351", + "issue:43874|issue:44261", + "issue:43531|issue:45341", + "issue:46752|issue:47116", + "issue:43097|issue:45588", + "issue:27957|issue:43352", + "issue:43582|issue:45341", + "issue:44117|issue:44987", + "issue:44263|issue:46172", + "issue:44484|issue:45478", + "issue:44360|issue:46172", + "issue:43097|issue:46050" + ], + "prepared_review_unit_hash": "cebcd8f15b839a35eadc69cbc36b9fb15f9e3ad25c988b0c8d9f412b6b04a57a", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36566, + "estimated_input_tokens": 9142, + "estimated_eval_tokens": 18540 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft pair looks like a true duplicate; the candidates are mostly distinct regressions or model-specific bugs across unrelated code paths. The strongest standalone issue is the v5 API-removal report because it is concrete, well-scoped, and broadly actionable.", + "confidence": 0.67, + "canonical_issue_reason": "issue:43097 is the clearest, most actionable report: a specific v5 regression about an API being removed without notice, with direct user impact and a well-defined fix target.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43097 is the best single issue to anchor this set because it is concrete, broadly relevant to v5 migration, and easier to triage/fix than the more model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API-settings guidance is unrelated to the MLA q_a_layernorm rms_norm_eps precision bug." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "TF32 behavior control and FineGrainedFP8 accuracy are different numeric issues with different code paths." + }, + { + "left": "issue:44261", + "right": "issue:46762", + "accept": false, + "reason": "Both mention numerical differences, but one is a missing eps/config bug and the other is a Minimax attention accuracy mismatch." + }, + { + "left": "issue:43208", + "right": "issue:46897", + "accept": false, + "reason": "xLSTM training blockers and Florence2 double-shift loss are unrelated model-specific bugs." + }, + { + "left": "issue:44077", + "right": "issue:46981", + "accept": false, + "reason": "Optional post_init validation and HF_ENDPOINT retry behavior are different subsystems with no shared bug path." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "A Qwen3-VL load shape mismatch is not the same as missing attn_qkv/attn_gate projections in qwen3next." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "Default shard-size behavior is unrelated to the Qwen3.5 MoE from_pretrained failure." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "SGD argument handling and deepspeed causing double optimizer steps are different optimizer bugs." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Wrong tokenizer decoder type and changed tokenization output are related only at a high level, not the same concrete bug." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Missing projections in qwen3next and Qwen3.5 MoE from_pretrained errors affect different model families and failure modes." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "A torch.split return-value issue in the DSA indexer is not the same as the GLM-5 RoPE implementation bug." + }, + { + "left": "issue:29925", + "right": "issue:30947", + "accept": false, + "reason": "SigLIP fast-tokenizer enforcement and LLaMA3 BOS/EOS insertion are separate tokenizer bugs." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Chunked-prefill cache handling and a testing_utils bug have no shared code path." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "An API removal warning and a gguf_file loading breakage are different from_pretrained issues, not one duplicate bug." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Both are Qwen3-VL-related, but one is a weight-shape loading mismatch and the other is a broader v5/MoE bug report." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "PretrainedConfig type checking and the tensorboard-without-TensorFlow backend error are unrelated regressions." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "A generic Qwen3-VL-Moe v5 bug and fp8 inference support are not the same fixable change." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "FlashAttention model-load failure and flash_attention_3 degenerate training are distinct failures in different code paths." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "The bfloat16 compile warning and BLIP2 dtype remaining float32 are different dtype-handling bugs." + }, + { + "left": "issue:43957", + "right": "issue:44351", + "accept": false, + "reason": "Meta-device model loading breakage and missing HybridCache import are unrelated v5 API/runtime issues." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Missing image-processor method and MLA q_a_layernorm precision error are different multimodal vs numeric bugs." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and a testing_utils bug are unrelated." + }, + { + "left": "issue:46752", + "right": "issue:47116", + "accept": false, + "reason": "Empty-list chat-template IndexError and streaming pipeline regression are separate failures." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "API-removal visibility and flash_attention.py crashing on s_aux=None are different issues in different parts of the stack." + }, + { + "left": "issue:27957", + "right": "issue:43352", + "accept": false, + "reason": "Flash Attention 2 support for XLMRoberta and NemotronH are model-specific support requests, not duplicates." + }, + { + "left": "issue:43582", + "right": "issue:45341", + "accept": false, + "reason": "AppleSilicon allocator warmup TypeError and testing_utils are unrelated." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES returning None and a model-loading failure in transformers>=5.1.0 are different from_pretrained problems." + }, + { + "left": "issue:44263", + "right": "issue:46172", + "accept": false, + "reason": "GlmMoeDsaIndexer torch.split behavior and deepseek-v4 comb transpose logic are unrelated implementation bugs." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "max_shard_size default discussion and the Qwen3.5 MoE from_pretrained error are not the same issue." + }, + { + "left": "issue:44360", + "right": "issue:46172", + "accept": false, + "reason": "DSA missing ReLU and deepseek-v4 transpose/dtype behavior are different code paths." + }, + { + "left": "issue:43097", + "right": "issue:46050", + "accept": false, + "reason": "An API removal warning and a sinq save/load quantization breakage are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 60, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:29159", + "issue:30947", + "issue:33453", + "issue:34567", + "issue:38617", + "issue:41950", + "issue:42757", + "issue:42890", + "issue:43012", + "issue:43064", + "issue:43097", + "issue:43208", + "issue:43550", + "issue:43673", + "issue:43873", + "issue:43874", + "issue:43986", + "issue:44038", + "issue:44062", + "issue:44117", + "issue:44246", + "issue:44261", + "issue:44263", + "issue:44315", + "issue:44410", + "issue:44484", + "issue:44485", + "issue:44704", + "issue:44936", + "issue:45103", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45440", + "issue:45753", + "issue:46123", + "issue:46172", + "issue:46298", + "issue:46365", + "issue:46369", + "issue:46655", + "issue:46664", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46821", + "issue:46981" + ], + "soft_pairs": [ + "issue:43986|issue:45372", + "issue:44261|issue:44485", + "issue:29159|issue:30947", + "issue:43097|issue:43873", + "issue:43208|issue:43550", + "issue:43986|issue:45341", + "issue:45753|issue:46369", + "issue:45440|issue:46172", + "issue:44246|issue:46981", + "issue:44117|issue:46701", + "issue:43012|issue:46821", + "issue:44410|issue:45375", + "issue:44062|issue:44117", + "issue:30947|issue:33453", + "issue:43673|issue:46123", + "issue:30947|issue:34567", + "issue:43874|issue:44263", + "issue:44038|issue:44410", + "issue:46655|issue:46664", + "issue:46172|issue:46710", + "issue:46298|issue:46752", + "issue:38617|issue:42757", + "issue:44704|issue:45103", + "issue:44936|issue:46365", + "issue:43064|issue:44315", + "issue:38617|issue:41950", + "issue:43986|issue:44484", + "issue:42890|issue:46298" + ], + "prepared_review_unit_hash": "afe1983dae7fc6d4d969f7af58e7d9b2e92b0cc05e73f71f58ac3506e39d9b95", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35286, + "estimated_input_tokens": 8822, + "estimated_eval_tokens": 17900 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a loose semantic cluster of mostly unrelated Transformers issues. The soft edges mostly connect items that share a subsystem name (tokenizers, processors, DeepSeek/Qwen variants, trainer, etc.) but not the same concrete bug or fix path, so none should be merged as duplicates.", + "confidence": 0.62, + "canonical_issue_reason": "Issue 33453 is the broadest representative of the cluster: a tokenizer-loading regression with wording that sits nearest to the other tokenizer-related reports. It is still not a duplicate of most others, but it is the best single anchor issue for this mixed cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "33453 is the best global issue candidate because it is generic, central to one of the recurring themes in the cluster, and broad enough to serve as the representative issue even though the cluster is not actually a tight duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Both are processor-loading import crashes, but they come from different missing dependencies and different code paths (video model/torchvision vs Gemma 4/mistral_common)." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "These describe different bugs: a q_a_layernorm precision/config issue versus a GLM-5 RoPE implementation question." + }, + { + "left": "issue:29159", + "right": "issue:30947", + "accept": false, + "reason": "Both are tokenizer-related, but one is slow/fast tokenizer inconsistency and the other is LLaMA3 special-token flags not taking effect; not the same concrete bug." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "One is about a removed model API flag, the other about offloading with quantization; different features and failure modes." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs and a Bamba torch.compile/SDPA failure are unrelated model-specific problems." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "A missing torchvision import during processor loading is unrelated to a small testing_utils bug." + }, + { + "left": "issue:45753", + "right": "issue:46369", + "accept": false, + "reason": "These affect different models and symptoms: an infinite loop on one image versus a memory overflow in sam3-litetext." + }, + { + "left": "issue:45440", + "right": "issue:46172", + "accept": false, + "reason": "Both mention DeepSeek, but one is a native-vs-remote implementation divergence and the other is a dtype/transpose bug in a different model path." + }, + { + "left": "issue:44246", + "right": "issue:46981", + "accept": false, + "reason": "Import slowness is not the same as repeated 429 retry behavior in AutoProcessor with HF_ENDPOINT." + }, + { + "left": "issue:44117", + "right": "issue:46701", + "accept": false, + "reason": "A TOKENIZER_MAPPING_NAMES None-handling bug is unrelated to the DiffusionGemma image-text pipeline failure." + }, + { + "left": "issue:43012", + "right": "issue:46821", + "accept": false, + "reason": "These are different issues: a PyTorch precision warning during compilation versus an AttributeError on TimesFM window_size." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Missing projections in Qwen3Next and a silently dropped config field in Qwen3_5MoeVisionConfig are different model/config bugs." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "AddedToken keyword duplication and TOKENIZER_MAPPING_NAMES returning None are different tokenizer-layer failures." + }, + { + "left": "issue:30947", + "right": "issue:33453", + "accept": false, + "reason": "The first is a LLaMA3 add_bos/add_eos behavior issue; the second is a broader tokenizer-loading regression. Related area, but not the same bug." + }, + { + "left": "issue:43673", + "right": "issue:46123", + "accept": false, + "reason": "Generation cache missing in chunked_prefill and mask pipeline batch-finalization are different pipeline logic bugs." + }, + { + "left": "issue:30947", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer special-token behavior and TrainerState token counting are unrelated." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Missing get_number_of_image_patches and torch.split return values are different GLM-related issues with different failing code paths." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Qwen3-VL-Moe and Qwen3Next are different model families with different bugs." + }, + { + "left": "issue:46655", + "right": "issue:46664", + "accept": false, + "reason": "Both are SmolVLM video-processor issues, but one is padding calculation and the other is num_frames semantics; they are distinct fixes." + }, + { + "left": "issue:46172", + "right": "issue:46710", + "accept": false, + "reason": "A tensor-operation bug in deepseek-v4 is not the same as an accuracy regression after upgrading Transformers for a different model." + }, + { + "left": "issue:46298", + "right": "issue:46752", + "accept": false, + "reason": "Noisy logging and an empty-list apply_chat_template crash are unrelated." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "These are different import errors from different external packages and symbols; not the same failure." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and an auto_docstring annotations crash are unrelated code paths." + }, + { + "left": "issue:44936", + "right": "issue:46365", + "accept": false, + "reason": "trainer.evaluate() after train() and Trainer import blocked by datasets are different trainer-related problems." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "A distributed optimizer-state bug and Liger Kernel not being applied with model_init are different training/integration issues." + }, + { + "left": "issue:38617", + "right": "issue:41950", + "accept": false, + "reason": "An import error for layer_type_validation and a video-classification pipeline looking for image processors are unrelated." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "Processor-loading failures and a save_pretrained shard-size question are not the same issue." + }, + { + "left": "issue:42890", + "right": "issue:46298", + "accept": false, + "reason": "A flaky SAM HQ test due to missing set_seed and a noisy logging complaint are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 61, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:30947", + "issue:33357", + "issue:36010", + "issue:36683", + "issue:42175", + "issue:42757", + "issue:42831", + "issue:42907", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43352", + "issue:43502", + "issue:43526", + "issue:43550", + "issue:43582", + "issue:43756", + "issue:43824", + "issue:43825", + "issue:43827", + "issue:43906", + "issue:43976", + "issue:43992", + "issue:44077", + "issue:44162", + "issue:44261", + "issue:44368", + "issue:44380", + "issue:44451", + "issue:44484", + "issue:44485", + "issue:44655", + "issue:44734", + "issue:44938", + "issue:45479", + "issue:45584", + "issue:45712", + "issue:45753", + "issue:46132", + "issue:46172", + "issue:46178", + "issue:46298", + "issue:46365", + "issue:46421", + "issue:47116", + "issue:47276" + ], + "soft_pairs": [ + "issue:43502|issue:43992", + "issue:44380|issue:46421", + "issue:44451|issue:47116", + "issue:44938|issue:46178", + "issue:43208|issue:43526", + "issue:43064|issue:47116", + "issue:43012|issue:43550", + "issue:45712|issue:46298", + "issue:42907|issue:45753", + "issue:43066|issue:43906", + "issue:42907|issue:44655", + "issue:45584|issue:46365", + "issue:43012|issue:44368", + "issue:43825|issue:44655", + "issue:44734|issue:47116", + "issue:43827|issue:47116", + "issue:43352|issue:43976", + "issue:43992|issue:44484", + "issue:44077|issue:46365", + "issue:30947|issue:33357", + "issue:44380|issue:46132", + "issue:36683|issue:43582", + "issue:44380|issue:45479", + "issue:44162|issue:46172", + "issue:42757|issue:43824", + "issue:42175|issue:46365", + "issue:43012|issue:47276", + "issue:43756|issue:44485", + "issue:43352|issue:44380", + "issue:43097|issue:46298", + "issue:36010|issue:42757", + "issue:42831|issue:44261" + ], + "prepared_review_unit_hash": "34b1ecfd656c4f4da239869e9bedc18e670e348ab8a3f5040fc62ce174f2c42d", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37101, + "estimated_input_tokens": 9276, + "estimated_eval_tokens": 18808 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly unrelated issues with only superficial topical overlap; none of the soft pairs look like true duplicates or the same change. If forced to pick a representative issue, issue 43097 is the clearest standalone regression report and has inbound references, but this set is too diverse to treat as one cluster.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 43097 is the most central, well-scoped regression report in this set and has inbound references; it is the best single representative only because the rest are clearly separate topics.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43097 has the strongest triage value as a concrete, user-visible v5 regression with broader discussion/reference signals than the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "Different failures: network/API access despite local_files_only vs missing embed_tokens.weight in from_pretrained." + }, + { + "left": "issue:44380", + "right": "issue:46421", + "accept": false, + "reason": "GPT2 attention scaling being ignored is unrelated to cache lazy initialization causing recompilation." + }, + { + "left": "issue:44451", + "right": "issue:47116", + "accept": false, + "reason": "Model loading failure for a specific checkpoint is not the same bug as streaming pipeline regression in v5." + }, + { + "left": "issue:44938", + "right": "issue:46178", + "accept": false, + "reason": "Python 3.14 import/load failure and tensorflow_text backend mapping error are different import-path problems." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training bugs and BeitImageProcessorFast reduce_labels returning one label are unrelated." + }, + { + "left": "issue:43064", + "right": "issue:47116", + "accept": false, + "reason": "FSDP2/PEFT optimizer state corruption is a different code path from streaming pipeline behavior." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A bfloat16 precision warning is not the same as Bamba SDPA torch.compile failure." + }, + { + "left": "issue:45712", + "right": "issue:46298", + "accept": false, + "reason": "Dummy-class cleanup/check_repo leakage is unrelated to a noisy logging message." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Saving dequantized models and an infinite-loop multimodal generation bug are different problems." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Wrong tokenizer decoder type is not the same as an isolated reproduction of another issue." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline save_pretrained failure is a different saving path than dequantized Ministrals/Devstrals save issues." + }, + { + "left": "issue:45584", + "right": "issue:46365", + "accept": false, + "reason": "Whisper empty-transcription generation bug is unrelated to Trainer import failure caused by datasets." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "A compile-time float32 warning is not the same as a tie_word_embeddings warning." + }, + { + "left": "issue:43825", + "right": "issue:47116", + "accept": false, + "reason": "Incorrect translation-task error text is separate from the streaming pipeline regression." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Unsupported Flash Attention 2 for NemotronH is unrelated to Python 3.9/3.10 compatibility." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing embed_tokens.weight on load is not the same as asking why max_shard_size defaults to 50GB." + }, + { + "left": "issue:44077", + "right": "issue:46365", + "accept": false, + "reason": "Optional post_init handling in patchtsmixer is unrelated to the Trainer import issue." + }, + { + "left": "issue:30947", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer BOS/EOS handling does not match a CLIP bus error on macOS." + }, + { + "left": "issue:44380", + "right": "issue:46132", + "accept": false, + "reason": "Ignored GPT2 attention scaling and altered registered-function behavior are different bugs." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Missing vocab_size on Gemma3Config is unrelated to an Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention scaling being ignored is not the same as degenerate zero loss in sequence classification." + }, + { + "left": "issue:44162", + "right": "issue:46172", + "accept": false, + "reason": "ESM2 breakage and deepseek-v4 tensor transpose behavior are separate model-specific issues." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Two different import errors for different symbols and packages; not the same underlying bug." + }, + { + "left": "issue:42175", + "right": "issue:46365", + "accept": false, + "reason": "Missing TensorFlow backend after pip install .[torch] is unrelated to Trainer import failure with datasets installed." + }, + { + "left": "issue:43012", + "right": "issue:47276", + "accept": false, + "reason": "A precision warning during compile and a misleading synced_gpus warning are different messages/issues." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Smollm3 RoPE layer dropping and GLM-5 RoPE implementation discussion are not the same concrete bug." + }, + { + "left": "issue:43097", + "right": "issue:46298", + "accept": false, + "reason": "An API removal regression is unrelated to a noisy logging message." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Importing GenerationMixin and importing is_offline_mode are different missing-symbol regressions." + }, + { + "left": "issue:42831", + "right": "issue:44261", + "accept": false, + "reason": "FineGrainedFP8 accuracy issues and MLA q_a_layernorm rms_norm_eps precision errors are distinct." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 62, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:29870", + "issue:29925", + "issue:33453", + "issue:36296", + "issue:36683", + "issue:38617", + "issue:42890", + "issue:43064", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43502", + "issue:43611", + "issue:43618", + "issue:43761", + "issue:43867", + "issue:43976", + "issue:43992", + "issue:44077", + "issue:44222", + "issue:44263", + "issue:44355", + "issue:44380", + "issue:44393", + "issue:44483", + "issue:44514", + "issue:44625", + "issue:44655", + "issue:44855", + "issue:44908", + "issue:44938", + "issue:45216", + "issue:45276", + "issue:45478", + "issue:45593", + "issue:45759", + "issue:46077", + "issue:46202", + "issue:46365", + "issue:46512", + "issue:46851", + "issue:46877", + "issue:47059", + "issue:47110", + "issue:47317", + "issue:47321", + "issue:47332", + "issue:47337" + ], + "soft_pairs": [ + "issue:44222|issue:45478", + "issue:44483|issue:44514", + "issue:43064|issue:44380", + "issue:42890|issue:47059", + "issue:43618|issue:43761", + "issue:45276|issue:46077", + "issue:44625|issue:46365", + "issue:38617|issue:43502", + "issue:44625|issue:46077", + "issue:44655|issue:47321", + "issue:44655|issue:47332", + "issue:43299|issue:43867", + "issue:43611|issue:44655", + "issue:43618|issue:44380", + "issue:36683|issue:43475", + "issue:43352|issue:45593", + "issue:36683|issue:44077", + "issue:46077|issue:46512", + "issue:43976|issue:44938", + "issue:44263|issue:45759", + "issue:44355|issue:46877", + "issue:29925|issue:33453", + "issue:43867|issue:45216", + "issue:44938|issue:47059", + "issue:36683|issue:44625", + "issue:43618|issue:45216", + "issue:36683|issue:47110", + "issue:29870|issue:36296", + "issue:44355|issue:44855", + "issue:43867|issue:46202", + "issue:44908|issue:46851", + "issue:43992|issue:47317", + "issue:44393|issue:47337" + ], + "prepared_review_unit_hash": "1a6c064ff108596663784c1b900da04f6ce0c897d4301dc5dfcb4ea7ebc86b4a", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37332, + "estimated_input_tokens": 9333, + "estimated_eval_tokens": 18922 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose mix of mostly unrelated Transformers issues. The soft pairs share themes like loading, saving, CLIP, Gemma, or Python-version regressions, but they do not look like the same concrete bug/change.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43611 is the broadest representative here: a Transformers 5.0 loading regression affecting models with `base_model_prefix`, which is closer to the cluster\u2019s recurring load/save/config breakage theme than the more narrowly scoped reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Same reason: it is the most general, central loading-regression issue among an otherwise heterogeneous set, so it is the best single issue to anchor the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5/MoE and FP8/save-load behavior, but one is about saving pretrained Moe weights and the other about `from_pretrained` loading failure; too different to merge." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Different layers and failures: chat-completions request validation vs batched processor chat template padding behavior." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption is unrelated to GPT2 attention scaling under SDPA/FlashAttention." + }, + { + "left": "issue:42890", + "right": "issue:47059", + "accept": false, + "reason": "A flaky test seeding issue is not the same as a CLI completion compatibility break." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both are CLIP v5 regressions, but one is missing attentions and the other is missing hidden_states; related area, not clearly the same bug." + }, + { + "left": "issue:45276", + "right": "issue:46077", + "accept": false, + "reason": "Gemma4 embedding resize behavior and Gemma4 bidirectional-attention masking are separate code paths." + }, + { + "left": "issue:44625", + "right": "issue:46365", + "accept": false, + "reason": "Qwen3.5 `num_labels` propagation and an import conflict with `datasets` are unrelated." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "An import error for `layer_type_validation` is unrelated to `local_files_only=True` still triggering network access." + }, + { + "left": "issue:44625", + "right": "issue:46077", + "accept": false, + "reason": "Same broad Gemma/Qwen family style, but the bugs are distinct config-vs-attention issues." + }, + { + "left": "issue:44655", + "right": "issue:47321", + "accept": false, + "reason": "Saving `Pipeline` objects and an AMD quark class update problem are different." + }, + { + "left": "issue:44655", + "right": "issue:47332", + "accept": false, + "reason": "Both mention saving, but one is pipeline serialization and the other is disk-offloaded Qwen3-VL saving." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Qwen3VL MoE loading regression and a sorted-state-dict load failure are not the same concrete problem." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "Model loading with `base_model_prefix` and pipeline save_pretrained are distinct issues." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "CLIP output-field regression and GPT2 attention scaling under backend selection are unrelated bugs." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Missing `vocab_size` in Gemma3Config is unrelated to missing `fpn_position_embeddings` in SAM 3 Video." + }, + { + "left": "issue:43352", + "right": "issue:45593", + "accept": false, + "reason": "Flash Attention 2.0 support gating is not the same bug as D-FINE auxiliary-loss handling." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "A missing config attribute and the `patchtsmixer` optional `post_init` restriction are unrelated." + }, + { + "left": "issue:46077", + "right": "issue:46512", + "accept": false, + "reason": "Bidirectional attention mask construction and stale `config.dtype` after `from_config(dtype=...)` are different issues." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python-version compatibility reports, but they target different versions and different failures." + }, + { + "left": "issue:44263", + "right": "issue:45759", + "accept": false, + "reason": "A `torch.split()` return-value issue in GlmMoeDsaIndexer is unrelated to Qwen composite-config unwrapping in `from_config`." + }, + { + "left": "issue:44355", + "right": "issue:46877", + "accept": false, + "reason": "Compiled-Python-file errors and noisy auto-docstring lint output are unrelated." + }, + { + "left": "issue:29925", + "right": "issue:33453", + "accept": false, + "reason": "Tokenizer fast-mode enforcement and a general tokenizer-loading regression are too broad to treat as the same bug." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Sorted state-dict loading errors and Qwen3.5 checkpoint saving corruption are different directions of failure." + }, + { + "left": "issue:44938", + "right": "issue:47059", + "accept": false, + "reason": "Python 3.14 import/load failure and the huggingface_hub CLI completion regression are unrelated." + }, + { + "left": "issue:36683", + "right": "issue:44625", + "accept": false, + "reason": "Gemma3 config `vocab_size` and Qwen3.5 `num_labels` propagation are separate config bugs." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIP attentions regression and Qwen3.5 checkpoint save corruption do not share the same code path." + }, + { + "left": "issue:36683", + "right": "issue:47110", + "accept": false, + "reason": "Missing `vocab_size` and `special_tokens` being a list instead of a dict are unrelated." + }, + { + "left": "issue:29870", + "right": "issue:36296", + "accept": false, + "reason": "Bart generation masking error and tensor-parallel training state issues are different subsystems." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Compiled-file execution errors and a Python 3.13 `@torch.jit.script` parsing/indentation issue are distinct." + }, + { + "left": "issue:43867", + "right": "issue:46202", + "accept": false, + "reason": "A sorted state-dict loading bug is not the same as Gemma pipeline roundtrip failure on the `\u2581` token." + }, + { + "left": "issue:44908", + "right": "issue:46851", + "accept": false, + "reason": "Scheduler kwarg handling and Qwen3.5 variable-length `position_ids` behavior are unrelated." + }, + { + "left": "issue:43992", + "right": "issue:47317", + "accept": false, + "reason": "Missing `embed_tokens.weight` in UMT5Encoder has nothing to do with PPFormulaNet training loss." + }, + { + "left": "issue:44393", + "right": "issue:47337", + "accept": false, + "reason": "Qwen3-VL 2D bbox output issues and a flaky bbox-range test are not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 63, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28282", + "issue:29870", + "issue:33453", + "issue:36296", + "issue:41628", + "issue:41669", + "issue:42548", + "issue:42890", + "issue:43010", + "issue:43064", + "issue:43065", + "issue:43502", + "issue:43525", + "issue:43575", + "issue:43618", + "issue:43723", + "issue:43867", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44075", + "issue:44077", + "issue:44112", + "issue:44222", + "issue:44230", + "issue:44246", + "issue:44351", + "issue:44360", + "issue:44387", + "issue:44393", + "issue:44492", + "issue:44556", + "issue:44734", + "issue:44964", + "issue:44977", + "issue:45125", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45375", + "issue:45440", + "issue:45478", + "issue:45561", + "issue:45850", + "issue:46365", + "issue:46617", + "issue:47000", + "issue:47317" + ], + "soft_pairs": [ + "issue:44075|issue:46365", + "issue:43010|issue:45278", + "issue:44393|issue:47000", + "issue:42548|issue:44351", + "issue:43992|issue:45310", + "issue:29870|issue:33453", + "issue:43064|issue:44492", + "issue:36296|issue:43575", + "issue:44222|issue:45125", + "issue:43618|issue:44077", + "issue:41669|issue:44246", + "issue:42890|issue:45561", + "issue:43867|issue:45357", + "issue:43867|issue:45375", + "issue:28282|issue:33453", + "issue:43010|issue:43976", + "issue:43618|issue:44977", + "issue:43010|issue:45850", + "issue:45125|issue:45478", + "issue:45125|issue:45375", + "issue:44360|issue:45440", + "issue:43502|issue:44062", + "issue:42890|issue:44112", + "issue:43525|issue:47317", + "issue:42890|issue:44964", + "issue:44734|issue:46617", + "issue:44387|issue:44556", + "issue:41628|issue:42548", + "issue:43723|issue:44556", + "issue:44222|issue:44230", + "issue:43010|issue:43065" + ], + "prepared_review_unit_hash": "664ae10f597cdda87825de472784c84afb36186f0ce020128a82d0910c084a98", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36425, + "estimated_input_tokens": 9107, + "estimated_eval_tokens": 18470 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of distinct bugs, not one duplicate family. The only clear duplicate-like pair is the import-time regression reports in `issue:41669` and `issue:44246`; most other soft edges are too broad or point to different code paths, models, or failure modes.", + "confidence": 0.87, + "canonical_issue_reason": "`issue:41669` is the most concrete and root-cause-oriented issue in the cluster: it identifies a specific import-time regression (`import *` in model modules) and proposes a precise fix, which also aligns with the nearby slow-import report in `issue:44246`.", + "canonical_pr_reason": null, + "best_issue_reason": "`issue:41669` is the best representative issue because it is specific, actionable, and anchored to a clear code-path regression rather than a broad symptom report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44075", + "right": "issue:46365", + "accept": false, + "reason": "Different problems: optimizer argument handling vs an import failure involving `Trainer` and `datasets`." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "Both mention regressions, but one is about cache update methods and the other is broad import errors after upgrade; not the same bug." + }, + { + "left": "issue:44393", + "right": "issue:47000", + "accept": false, + "reason": "Different model/task failures: Qwen3-VL bounding-box output issue vs RTDetr loss infeasibility." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and likely different root causes." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Different `from_pretrained` failures on different models; no evidence they share one concrete code-path bug." + }, + { + "left": "issue:29870", + "right": "issue:33453", + "accept": false, + "reason": "Bart generation UnboundLocalError and tokenizer-loading regression are unrelated." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "Unrelated: FSDP/PEFT optimizer-state bug versus a cache-strategy typo." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Both involve tensor parallelism, but one is a training bug and the other is an OOM on loading a specific large model." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "Different issues: FP8 `save_pretrained` behavior versus missing `_tp_plan` for tensor parallelism." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Completely different model/output regression versus an API restriction on `patchtsmixer` post-init." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": true, + "reason": "Both describe the same underlying regression: `import transformers` becoming slow, with `issue:41669` pointing to a specific root cause in model imports." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Different test failures: missing `set_seed()` in a model integration test vs xdist race on captured info output." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Both are model loading/saving problems, but they affect different mechanisms and are not the same code-path." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "A generic state-dict load error is not the same as a missing vision-config field silently dropped by strict parsing." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "PyTorch-not-found import error and tokenizer-loading regression are unrelated." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "Cache-update no-grad concern is unrelated to a Python version compatibility issue." + }, + { + "left": "issue:43618", + "right": "issue:44977", + "accept": false, + "reason": "Different failures in different models; no shared concrete bug." + }, + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "One is a specific cache-update bug; the other is a vague broad complaint about many models breaking." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "Related model family, but one is a missing tensor-parallel plan and the other is a generic `from_pretrained` error; too little to treat as the same bug." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Tensor-parallel plan missing and vision-config field dropping are different issues." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "Different domains entirely: DSA indexer implementation vs DeepSeekV3MoE parity with remote code." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "Offline-download behavior and AddedToken constructor argument collision are unrelated." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Separate flaky tests in different models and subsystems." + }, + { + "left": "issue:43525", + "right": "issue:47317", + "accept": false, + "reason": "Llama4 config attribute error and PPFormulaNet training-loss bug do not share a code path." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Different model-specific test flake vs model loading failure." + }, + { + "left": "issue:44734", + "right": "issue:46617", + "accept": false, + "reason": "KV-cache continuation crash and cached trust-remote-code file lookup failure are unrelated." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Different regressions: int4 quantization memory growth vs reloading older checkpoints in v5." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Both are import errors, but they concern different exported symbols and likely different breakages." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Tokenizer-loading issue and checkpoint reload incompatibility are distinct failures." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "FP8 save behavior and FP8 inference support are not the same bug." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Duplicate of an earlier reject: unrelated test failures." + }, + { + "left": "issue:43502", + "right": "issue:47317", + "accept": false, + "reason": "Offline-loading bug and PPFormulaNet training-loss bug are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The clustering summary is grounded in the packet and stays conservative: it treats most pairs as distinct issues unless the titles/body clearly align. The lone accepted edge, `issue:41669`/`issue:44246`, is reasonably supported by the shared import-slowdown symptom and the specific import-* root cause in 41669. No soft-edge verdict overstates evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 64, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:29870", + "issue:30333", + "issue:30990", + "issue:31795", + "issue:36296", + "issue:38617", + "issue:42757", + "issue:42898", + "issue:43010", + "issue:43232", + "issue:43299", + "issue:43475", + "issue:43502", + "issue:43576", + "issue:43824", + "issue:43827", + "issue:43874", + "issue:43901", + "issue:44038", + "issue:44077", + "issue:44222", + "issue:44230", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44485", + "issue:44556", + "issue:44623", + "issue:44749", + "issue:44861", + "issue:44961", + "issue:45071", + "issue:45125", + "issue:45230", + "issue:45362", + "issue:45375", + "issue:45850", + "issue:46095", + "issue:46132", + "issue:46143", + "issue:46144", + "issue:46617", + "issue:46664", + "issue:46726", + "issue:47059", + "issue:47110", + "issue:47317" + ], + "soft_pairs": [ + "issue:43824|issue:46617", + "issue:43576|issue:46132", + "issue:45071|issue:47317", + "issue:42898|issue:46664", + "issue:46143|issue:46144", + "issue:43502|issue:44485", + "issue:44355|issue:44749", + "issue:43874|issue:46617", + "issue:30990|issue:31795", + "issue:45850|issue:47110", + "issue:43010|issue:43232", + "issue:43502|issue:44222", + "issue:43299|issue:44556", + "issue:46664|issue:46726", + "issue:30333|issue:36296", + "issue:43827|issue:44961", + "issue:44230|issue:46095", + "issue:29870|issue:31795", + "issue:44038|issue:45125", + "issue:43576|issue:43824", + "issue:43502|issue:44961", + "issue:44355|issue:45071", + "issue:43576|issue:45071", + "issue:43576|issue:47059", + "issue:42757|issue:43576", + "issue:36296|issue:43901", + "issue:31795|issue:38617", + "issue:43475|issue:46617", + "issue:44077|issue:44861", + "issue:44623|issue:45230", + "issue:44351|issue:46617", + "issue:44315|issue:44360", + "issue:45362|issue:45850", + "issue:45375|issue:47110", + "issue:44485|issue:44961" + ], + "prepared_review_unit_hash": "a0e5c8d3e568b820b2920f7bf773bc04a7f02c77f520bda4c2dc8c2d299d6083", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37721, + "estimated_input_tokens": 9431, + "estimated_eval_tokens": 19118 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is dominated by loose textual similarity around Transformers v5 regressions, but the suggested pairs do not describe the same underlying bug/change. I would not merge any of the soft edges.", + "confidence": 0.95, + "canonical_issue_reason": "Issue 43576 is the best representative of the loose cluster because it is a concrete v5 regression and acts as a hub in the similarity graph, but it is still not a true duplicate anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "43576 is the clearest, most concrete issue here; the rest are mostly unrelated breakages, docs problems, or different model-specific failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43824", + "right": "issue:46617", + "accept": false, + "reason": "Different failures: one is a missing Qwen import, the other is a cache FileNotFoundError for trust_remote_code." + }, + { + "left": "issue:43576", + "right": "issue:46132", + "accept": false, + "reason": "One is a broken CLI/env command in v5; the other is an AttentionInterface.register behavior change." + }, + { + "left": "issue:45071", + "right": "issue:47317", + "accept": false, + "reason": "PretrainedConfig type-checking and PPFormulaNet training loss are unrelated problems." + }, + { + "left": "issue:42898", + "right": "issue:46664", + "accept": false, + "reason": "Tokenization cleanup behavior and SmolVLM video frame semantics are different subsystems and bugs." + }, + { + "left": "issue:46143", + "right": "issue:46144", + "accept": false, + "reason": "Both involve RoFormer, but one is kwargs propagation and the other is attention-interface usage; not the same bug." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "local_files_only network leakage and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors and a performance regression after upgrade are not the same issue." + }, + { + "left": "issue:43874", + "right": "issue:46617", + "accept": false, + "reason": "A missing image-processor method is unrelated to failing to load a trust_remote_code model from cache." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Loading hang for Sentence Transformers and documentation confusion in model.forward are different topics." + }, + { + "left": "issue:45850", + "right": "issue:47110", + "accept": false, + "reason": "Broad v5 complaints do not match a specific special_tokens AttributeError." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "Both touch generation internals, but one is about no_grad decoration and the other about sync_gpus model-kwargs updates." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "local_files_only API requests and FP8 save_pretrained for moe are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Qwen3-VL-Moe loading and reloading old checkpoints after upgrade are different concrete failures." + }, + { + "left": "issue:46664", + "right": "issue:46726", + "accept": false, + "reason": "SmolVLM video-frame semantics and missing vision backend handling are distinct bugs." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLflow reporting on training failure and tensor-parallel training bugs are unrelated." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs still mentioning pipeline() and the 'racoon' issue are not the same change or bug." + }, + { + "left": "issue:44230", + "right": "issue:46095", + "accept": false, + "reason": "FP8 Qwen3-VL/moe support and DeepSeekV4 weight conversion are different requests." + }, + { + "left": "issue:29870", + "right": "issue:31795", + "accept": false, + "reason": "A generation UnboundLocalError in Bart and forward-argument documentation confusion are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "Both involve Qwen3.5/Moe wording, but one is a loading bug and the other is missing tensor-parallel plan support." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "CLI/env breakage and a specific import error for Qwen2.5-VL are different failures." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "local_files_only behavior and an unrelated 'racoon' issue do not match." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Compiled-file runtime errors and PretrainedConfig type checking are unrelated regressions." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "A broken env command is not the same as type-checking failures in PretrainedConfig." + }, + { + "left": "issue:43576", + "right": "issue:47059", + "accept": false, + "reason": "CLI/env breakage and a huggingface_hub completion attribute mismatch are separate issues." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "Missing is_offline_mode import and broken transformers env are different problems." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training and text-classification docs mentioning return_all_scores are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "A docs clarification issue and an ImportError for layer_type_validation are not duplicates." + }, + { + "left": "issue:43475", + "right": "issue:46617", + "accept": false, + "reason": "SAM 3 Video attribute loss and trust_remote_code cache loading failure are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "A patchtsmixer post_init policy question and a tied-weights AttributeError are different bugs." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "Missing processor.save_pretrained files and a generic bug report are not the same underlying issue." + }, + { + "left": "issue:44351", + "right": "issue:46617", + "accept": false, + "reason": "HybridCache import failure and cache FileNotFoundError are distinct failures." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Liger Kernel application with model_init and a DSA indexer ReLU discussion are unrelated." + }, + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "A Qwen3.5-35B chat crash is not the same as a broad complaint that v5 breaks many models." + }, + { + "left": "issue:45375", + "right": "issue:47110", + "accept": false, + "reason": "Missing Qwen3.5MoeVisionConfig field and special_tokens list-key access error are different bugs." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation and the 'racoon' issue do not match." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 65, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43298", + "issue:43352", + "issue:43408", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43867", + "issue:43976", + "issue:44246", + "issue:44273", + "issue:44295", + "issue:44297", + "issue:44315", + "issue:44351", + "issue:44393", + "issue:44466", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44655", + "issue:44704", + "issue:44861", + "issue:44908", + "issue:44961", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45230", + "issue:45341", + "issue:45381", + "issue:45468", + "issue:45542", + "issue:45753", + "issue:45850", + "issue:46018", + "issue:46095", + "issue:46172", + "issue:46209", + "issue:46218", + "issue:46328", + "issue:46664", + "issue:47059", + "issue:47110", + "issue:47176" + ], + "soft_pairs": [ + "issue:43673|issue:44492", + "issue:43352|issue:44961", + "issue:43525|issue:47110", + "issue:44273|issue:45542", + "issue:44961|issue:45468", + "issue:44315|issue:46209", + "issue:45381|issue:46664", + "issue:44273|issue:45003", + "issue:43867|issue:47110", + "issue:44466|issue:44556", + "issue:41669|issue:43673", + "issue:44655|issue:44861", + "issue:43572|issue:45850", + "issue:43525|issue:45850", + "issue:44485|issue:46664", + "issue:44393|issue:45753", + "issue:44393|issue:45071", + "issue:44295|issue:45230", + "issue:44556|issue:45092", + "issue:44297|issue:45230", + "issue:42371|issue:44492", + "issue:45850|issue:47059", + "issue:44246|issue:44273", + "issue:46018|issue:46172", + "issue:43408|issue:44315", + "issue:44492|issue:45850", + "issue:41669|issue:46218", + "issue:43976|issue:44393", + "issue:44492|issue:46328", + "issue:42890|issue:44908", + "issue:44704|issue:44908", + "issue:44908|issue:45071", + "issue:41669|issue:45341", + "issue:41669|issue:44351", + "issue:43298|issue:46095", + "issue:43576|issue:44908", + "issue:43576|issue:47176", + "issue:43976|issue:46218" + ], + "prepared_review_unit_hash": "93366884b147c05427d5686c215bcfefc1b3a7d6eda0b619ae176d3877e85632", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 38, + "serialized_chars": 38445, + "estimated_input_tokens": 9612, + "estimated_eval_tokens": 19480 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab-bag of mostly unrelated Transformers issues, many around v5 regressions and model-specific bugs. None of the soft pairs look like clear same-bug duplicates, so I reject all candidate edges.", + "confidence": 0.39, + "canonical_issue_reason": "No single duplicate hub stands out; issue 44556 is the broadest, most user-facing v5 checkpoint reload regression in the set and is the best representative issue.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44556 is the strongest representative because it captures a broad, concrete v5 compatibility failure around reloading checkpoints, unlike the more model-specific or narrowly scoped reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Chunked-prefill cache missing vs a cache-strategy typo; same vocabulary, different bug." + }, + { + "left": "issue:43352", + "right": "issue:44961", + "accept": false, + "reason": "Flash Attention support for Nemotron vs an unrelated placeholder issue title." + }, + { + "left": "issue:43525", + "right": "issue:47110", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config vs special_tokens dict/list handling; different code paths." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading regression vs TensorFlow backend detection error; unrelated symptoms and fixes." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "Placeholder issue vs Gemma-4 audio positional encoding bug; no shared underlying defect." + }, + { + "left": "issue:44315", + "right": "issue:46209", + "accept": false, + "reason": "Liger Kernel/model_init wiring vs FP8 inference on B200; different subsystems and failures." + }, + { + "left": "issue:45381", + "right": "issue:46664", + "accept": false, + "reason": "Qwen2.5-VL video position IDs vs SmolVLM num_frames semantics; both multimodal, but not the same bug." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Lazy loading/import behavior vs unsafe sys.modules access; different root causes." + }, + { + "left": "issue:43867", + "right": "issue:47110", + "accept": false, + "reason": "State-dict loading error vs special_tokens structure error; unrelated load/save paths." + }, + { + "left": "issue:44466", + "right": "issue:44556", + "accept": false, + "reason": "Serialization inconsistency in tied weights vs generic checkpoint reload incompatibility; related area, but not the same concrete bug." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-* cleanup/performance issue vs missing generation cache in chunked_prefill; different problems." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Pipeline save_pretrained failure vs _tied_weights_keys list crash; different save-time code paths." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "StableLmConfig pad_token_idx regression vs broad v5 breakage claim; too general and not the same bug." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "Llama4Config missing pad_token_id vs broad v5 breakage report; not a duplicate." + }, + { + "left": "issue:44485", + "right": "issue:46664", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion vs SmolVLM video-frame semantics; unrelated." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "2D bbox output issue in Qwen3-VL vs infinite loop on a specific image in Qwen3.5; different failure modes." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "Model output bug vs PretrainedConfig type-checking regression; unrelated." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "position_ids buffer access error vs generic 'Bug report' placeholder; no evidence of same bug." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "tokenizer.save_pretrained mismatch vs generic report placeholder; unrelated." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 configuration API change vs cache strategy typo; different areas entirely." + }, + { + "left": "issue:45850", + "right": "issue:47059", + "accept": false, + "reason": "Broad v5 regression complaint vs cli failure with huggingface_hub; no shared code-path." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import slowdown sometimes vs lazy loading not working properly; possibly adjacent, but not clearly the same bug." + }, + { + "left": "issue:46018", + "right": "issue:46172", + "accept": false, + "reason": "DeepSeek-V4 shared expert gating vs a tensor transpose issue; unrelated implementation bugs." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "sam3_video/sam3_tracker type warning vs Liger Kernel not applied with model_init; different components." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "Cache-strategy typo vs broad v5 breakage report; not the same underlying defect." + }, + { + "left": "issue:41669", + "right": "issue:46218", + "accept": false, + "reason": "Import-* performance cleanup vs module import failure for Qwen3_5ForCausalLM; unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python version compatibility issue vs Qwen3-VL bounding-box hallucination; unrelated." + }, + { + "left": "issue:44492", + "right": "issue:46328", + "accept": false, + "reason": "Typo in cache strategies vs SequenceFeatureExtractor.pad numpy/list performance issue; different code paths." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Missing set_seed in tests vs scheduler kwargs being ignored; unrelated test vs scheduler bug." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor kwargs not forwarded to cached_file vs inverse_sqrt scheduler kwargs ignored; both kwargs issues, but different APIs." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwargs bug vs PretrainedConfig type-checking regression; different subsystems." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Import-* cleanup/performance issue vs a small testing_utils bug; not the same bug." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-* cleanup/performance issue vs HybridCache import failure; unrelated." + }, + { + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Audio continuous batching support vs DeepSeek-V4 weight conversion question; different topics." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "transformers env CLI breakage vs scheduler kwargs bug; unrelated." + }, + { + "left": "issue:43576", + "right": "issue:47176", + "accept": false, + "reason": "Broken env command vs shell-operator allowlist bypass security bug; different command-line paths and different severity." + }, + { + "left": "issue:43976", + "right": "issue:46218", + "accept": false, + "reason": "Python version compatibility regression vs module import error for Qwen3_5ForCausalLM; unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 66, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:28348", + "issue:29870", + "issue:31515", + "issue:31795", + "issue:33453", + "issue:37428", + "issue:38617", + "issue:41669", + "issue:42371", + "issue:42913", + "issue:43066", + "issue:43298", + "issue:43352", + "issue:43502", + "issue:43519", + "issue:43576", + "issue:43606", + "issue:43618", + "issue:43704", + "issue:43976", + "issue:44230", + "issue:44261", + "issue:44273", + "issue:44315", + "issue:44351", + "issue:44360", + "issue:44485", + "issue:44492", + "issue:44623", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:44998", + "issue:45230", + "issue:45250", + "issue:45290", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45656", + "issue:45850", + "issue:46218", + "issue:46655", + "issue:46894", + "issue:47059", + "issue:47110", + "issue:47176", + "issue:47177" + ], + "soft_pairs": [ + "issue:45230|issue:45656", + "issue:43976|issue:47176", + "issue:31795|issue:33453", + "issue:43519|issue:46655", + "issue:44315|issue:44829", + "issue:44861|issue:47110", + "issue:45468|issue:46894", + "issue:42913|issue:43066", + "issue:44360|issue:45468", + "issue:44485|issue:47176", + "issue:43976|issue:47177", + "issue:43519|issue:44485", + "issue:31795|issue:37428", + "issue:43576|issue:44273", + "issue:44485|issue:47177", + "issue:44273|issue:44908", + "issue:43519|issue:44908", + "issue:44861|issue:47059", + "issue:41669|issue:43704", + "issue:43576|issue:46218", + "issue:44485|issue:45468", + "issue:43606|issue:45468", + "issue:42371|issue:43704", + "issue:43704|issue:44485", + "issue:45850|issue:47177", + "issue:45850|issue:47176", + "issue:43352|issue:47177", + "issue:43502|issue:43519", + "issue:43519|issue:44623", + "issue:28348|issue:45250", + "issue:44261|issue:45468", + "issue:45850|issue:46218", + "issue:43704|issue:45341", + "issue:43352|issue:46218", + "issue:44351|issue:46218", + "issue:44861|issue:45290", + "issue:44230|issue:44623", + "issue:28348|issue:43352", + "issue:44908|issue:44998", + "issue:44998|issue:47177", + "issue:44998|issue:47176", + "issue:44230|issue:45310", + "issue:44492|issue:44998", + "issue:43618|issue:44998", + "issue:31515|issue:38617", + "issue:29870|issue:31515", + "issue:43298|issue:44230" + ], + "prepared_review_unit_hash": "2fd0aa8e595b50afd22519a9f6cc183f077103b13d6432da86739ec5ad927ac5", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 47, + "serialized_chars": 40992, + "estimated_input_tokens": 10248, + "estimated_eval_tokens": 20752 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab bag of mostly unrelated issues. There is some loose overlap around Flash Attention 2.0, tokenizer/model-loading regressions, and a few processor bugs, but none of the soft pairs look like the same concrete underlying bug or change. I would not merge any of the suggested pairs.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 45250 is the broadest and most central Flash Attention 2.0 request in the set, so it is the best umbrella candidate among otherwise unrelated issues.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45250 is the most representative high-level issue here; it is generic enough to cover the closest related flash-attention requests without overfitting to a specific model.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45230", + "right": "issue:45656", + "accept": false, + "reason": "Generic bug-report placeholder versus a specific DeepSpeed optimizer-step bug; no shared code path." + }, + { + "left": "issue:43976", + "right": "issue:47176", + "accept": false, + "reason": "Python version incompatibility and shell allowlist bypass are unrelated problems." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Docs confusion about forward inputs versus a tokenizer loading regression; different issues." + }, + { + "left": "issue:43519", + "right": "issue:46655", + "accept": false, + "reason": "Both involve processor calculations, but for different models and different bugs (timestamp vs padding)." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel application in model_init is unrelated to flash_attention_3 training degeneration." + }, + { + "left": "issue:44861", + "right": "issue:47110", + "accept": false, + "reason": "Different AttributeErrors in different code paths; no evidence they are the same bug." + }, + { + "left": "issue:45468", + "right": "issue:46894", + "accept": false, + "reason": "Gemma audio positional encoding and CohereASR training loss shift are distinct model-specific bugs." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both are tokenizer v5-related, but one is a broad behavior difference and the other is a decoder-type bug; not clearly the same fix." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer ReLU discussion and Gemma audio positional encoding are unrelated." + }, + { + "left": "issue:44485", + "right": "issue:47176", + "accept": false, + "reason": "RoPE implementation and shell command-substitution bypass are entirely different areas." + }, + { + "left": "issue:43976", + "right": "issue:47177", + "accept": false, + "reason": "Python version support and tar/zip slip security issue do not match." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Processor timestamp math versus scheduler kwargs handling are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:37428", + "accept": false, + "reason": "Documentation confusion and import error for flash support helper are not the same defect." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env command and lazy loading failure are separate behaviors." + }, + { + "left": "issue:44485", + "right": "issue:47177", + "accept": false, + "reason": "RoPE implementation bug and archive path traversal are unrelated." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading and inverse_sqrt scheduler kwargs are different subsystems." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Processor timestamp calculation does not overlap with LR scheduler kwargs handling." + }, + { + "left": "issue:44861", + "right": "issue:47059", + "accept": false, + "reason": "Tied-weight key handling and CLI completion breakage are unrelated." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import-star cleanup and VRAM leak in Qwen3ForCausalLM are not the same bug." + }, + { + "left": "issue:43576", + "right": "issue:46218", + "accept": false, + "reason": "Broken env command and Qwen3_5ForCausalLM import failure are distinct issues." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "Different model implementations and different failure modes." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to Gemma audio positional encoding." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 settings guidance and VRAM leak are unrelated." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak and RoPE implementation bug are not the same code-path." + }, + { + "left": "issue:45850", + "right": "issue:47177", + "accept": false, + "reason": "General breakage report versus archive traversal security issue; no shared fix." + }, + { + "left": "issue:45850", + "right": "issue:47176", + "accept": false, + "reason": "Broad regression complaint and shell allowlist bypass are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:47177", + "accept": false, + "reason": "Flash Attention 2 support request for Nemotron and tar-slip security bug are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "Unexpected network access and timestamp calculation are different bugs." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Incorrect timestamp math and missing save_pretrained files are separate processor issues." + }, + { + "left": "issue:28348", + "right": "issue:45250", + "accept": false, + "reason": "Both mention Flash Attention 2.0, but one is GPT2-specific support while the other is a generic title; not enough to prove the same concrete change." + }, + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "Different numerical precision/config issues in different model families." + }, + { + "left": "issue:45850", + "right": "issue:46218", + "accept": false, + "reason": "General v5 breakage and Qwen3_5 import failure are not the same underlying defect." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "RoPE implementation and audio positional encoding are distinct model components." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload mismatch and Gemma audio positional encoding are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:43704", + "accept": false, + "reason": "FP8 support request and VRAM leak are different topics." + }, + { + "left": "issue:43704", + "right": "issue:45341", + "accept": false, + "reason": "VRAM leak and a testing_utils bug do not appear to share code." + }, + { + "left": "issue:43352", + "right": "issue:46218", + "accept": false, + "reason": "Flash Attention support missing and module import failure are different problems." + }, + { + "left": "issue:44485", + "right": "issue:47177", + "accept": false, + "reason": "RoPE implementation bug and archive traversal vulnerability are unrelated." + }, + { + "left": "issue:44829", + "right": "issue:45290", + "accept": false, + "reason": "Flash_attention_3 training degeneration and chat template tool-call crash are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 inference support and processor save_pretrained missing files are not the same issue." + }, + { + "left": "issue:28348", + "right": "issue:43352", + "accept": false, + "reason": "Both are Flash Attention 2.0 support requests, but for different model classes and not clearly one fix." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Scheduler kwargs bug and unrelated placeholder issue." + }, + { + "left": "issue:44998", + "right": "issue:47177", + "accept": false, + "reason": "Noise issue versus security vulnerability; no overlap." + }, + { + "left": "issue:44998", + "right": "issue:47176", + "accept": false, + "reason": "Noise issue versus shell-operator bypass vulnerability." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "FP8 support request and Qwen3.5 MoE from_pretrained error are different problems." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "Typo in cache strategies and unrelated placeholder issue." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression and unrelated placeholder issue." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Slow from_pretrained loading and import failure of layer_type_validation are unrelated." + }, + { + "left": "issue:29870", + "right": "issue:31515", + "accept": false, + "reason": "BART generate UnboundLocalError and slow checkpoint loading are different bugs." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Audio continuous batching support and FP8 VL/MoE support are different feature gaps." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 28, + "review_unit_index": 67, + "review_unit_count": 67, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-540", + "nodes": [ + "issue:30333", + "issue:31515", + "issue:31795", + "issue:36296", + "issue:37934", + "issue:42936", + "issue:43240", + "issue:43668", + "issue:43717", + "issue:43813", + "issue:44031", + "issue:46018", + "issue:46144" + ], + "soft_pairs": [ + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:31515|issue:31795", + "issue:43717|issue:46144", + "issue:42936|issue:46018", + "issue:43668|issue:44031", + "issue:42936|issue:43240", + "issue:43240|issue:46144", + "issue:42936|issue:44031", + "issue:42936|issue:46144", + "issue:42936|issue:43813", + "issue:43668|issue:43813", + "issue:43240|issue:43668", + "issue:37934|issue:43668", + "issue:42936|issue:43668" + ], + "prepared_review_unit_hash": "d20f50b41f79c41bd21f5bcf13cd297b3115d204b9dc367eede63343c88eaa18", + "component_budget": { + "node_count": 540, + "item_count": 540, + "soft_pair_count": 1995, + "serialized_chars": 1033772, + "estimated_input_tokens": 258443, + "estimated_eval_tokens": 517142 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 15, + "serialized_chars": 11766, + "estimated_input_tokens": 2942, + "estimated_eval_tokens": 6140 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose collection of unrelated issue reports across training, loading, docs, tokenization, model architecture, and configuration. None of the soft pairs look like the same underlying bug or change, so I would not merge any of them as duplicates.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 43240 is the most concrete and broadly actionable report in the set, with the most inbound references; if one issue must anchor this cluster, it is the best fit.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43240 is the strongest representative: it describes a clear code-path bug rather than a typo/docs-only concern, and it appears to have the most external attention.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLflow job-status reporting bug vs. slow from_pretrained checkpoint loading; different subsystems and symptoms." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Checkpoint load performance issue vs tensor-parallel training bug; not the same failure mode." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Loading speed complaint vs confusing forward docs; one is runtime performance, the other documentation." + }, + { + "left": "issue:43717", + "right": "issue:46144", + "accept": false, + "reason": "Mamba/Mamba-2 init_weights behavior vs RoFormer attention interface usage; unrelated model implementations and code paths." + }, + { + "left": "issue:42936", + "right": "issue:46018", + "accept": false, + "reason": "Mask2former ignore_value handling vs DeepSeek-V4 shared expert gating; different models and bugs." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "Incorrect type hint in ModernBERTConfig vs tokenizer regex warning after version change; unrelated issues." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Both concern model/loss plumbing, but one is a specific ignore_value usage bug and the other is kwargs not being forwarded to loss calculation; not the same underlying bug." + }, + { + "left": "issue:43240", + "right": "issue:46144", + "accept": false, + "reason": "Loss-kwargs forwarding bug vs RoFormer attention-interface issue; different code paths." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Mask2former loss parameter bug vs tokenizer regex warning; no overlap in component or failure." + }, + { + "left": "issue:42936", + "right": "issue:46144", + "accept": false, + "reason": "Different models and different behaviors: ignore_value handling vs attention interface usage." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "Mask2former loss behavior vs a typo in quantization_operation text; unrelated." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "ModernBERT type-hint mistake vs quantization typo; both are minor metadata/text issues but not the same one." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Loss kwargs forwarding bug vs configuration type-hint error; separate bug classes." + }, + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Question about Llama4TextL2Norm semantics vs ModernBERT config typing; unrelated model details." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "Model loss parameter handling vs config type annotation; not the same issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10519-3", + "nodes": [ + "issue:10519", + "issue:11005", + "issue:11307" + ], + "soft_pairs": [ + "issue:10519|issue:11307", + "issue:10519|issue:11005", + "issue:11005|issue:11307" + ], + "prepared_review_unit_hash": "47bb4026a7f0071b2f0c56946b80bdcff6b639e434cf50ec83bc65c0b3b5df0c", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These three issues are not duplicates: one is about tokenizer truncation behavior, one about ReduceLROnPlateau-style scheduling, and one about Wav2Vec2 word offset extraction. They do not describe the same underlying bug or feature request.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue fits this cluster; the issues cover unrelated areas and solve different problems.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a good global representative because the cluster is heterogeneous rather than a duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:10519", + "right": "issue:11307", + "accept": false, + "reason": "Tokenizer truncation options and Wav2Vec2 word time offsets are unrelated feature requests with different code paths." + }, + { + "left": "issue:10519", + "right": "issue:11005", + "accept": false, + "reason": "Truncation-from-beginning and ReduceLROnPlateau-like scheduling address completely different components." + }, + { + "left": "issue:11005", + "right": "issue:11307", + "accept": false, + "reason": "Learning-rate scheduling functionality and speech word offset timing are distinct, non-overlapping issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 3, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-13244-3", + "nodes": [ + "issue:13244", + "issue:15980", + "issue:16618" + ], + "soft_pairs": [ + "issue:13244|issue:16618", + "issue:15980|issue:16618" + ], + "prepared_review_unit_hash": "8552eca202f87740a5af003b4072b6d2bfd5d3f2b4818dd2c13143fd59ac31af", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These three issues are all in the Transformers ecosystem, but they describe different problems: TAPAS tokenization mismatch, private-model auth/error messaging, and a generic BERT model-loading failure. I would not merge any pair as duplicates; issue 16618 is the broadest representative of the cluster\u2019s loose model-loading theme.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 16618 is the most general of the three and best matches the cluster\u2019s broad model-loading theme; the other two are narrower and clearly about different failure modes.", + "canonical_pr_reason": null, + "best_issue_reason": "16618 is the strongest representative because it is a user-facing model load failure, whereas 13244 is tokenizer-specific and 15980 is specifically about private-model download/auth messaging.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Reject: TAPAS tokenization differences are a tokenizer/implementation issue, while 16618 is a generic failure to load a BERT model. They are not the same underlying bug." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "Reject: 15980 is about an incorrect error message when downloading a private model without login; 16618 is a separate model-loading failure for a public model. Different cause and fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 4, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-19631-8", + "nodes": [ + "issue:19631", + "issue:19865", + "issue:20334", + "issue:20362", + "issue:22352", + "issue:22372", + "issue:25725", + "issue:25839" + ], + "soft_pairs": [ + "issue:19631|issue:22372", + "issue:19631|issue:20362", + "issue:19865|issue:25725", + "issue:19865|issue:22372", + "issue:20334|issue:22352", + "issue:20334|issue:20362", + "issue:20334|issue:25725", + "issue:20362|issue:25839", + "issue:20362|issue:25725", + "issue:22352|issue:25725", + "issue:22372|issue:25725" + ], + "prepared_review_unit_hash": "28fd20179938c96502f9b30ebfa6ddbc7a036675d008116ab7939a4916426988", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 11, + "serialized_chars": 8195, + "estimated_input_tokens": 2049, + "estimated_eval_tokens": 4354 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 11, + "serialized_chars": 8195, + "estimated_input_tokens": 2049, + "estimated_eval_tokens": 4354 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a false positive: the issues are about different model additions or feature requests, not the same underlying bug/change. None of the soft edges should be merged.", + "confidence": 0.97, + "canonical_issue_reason": "No true canonical issue exists here; the items request different models/features (EDSR/MDSR, VATT, FlexiBERT, Restormer, Meta-Transformer, D_Nikud, etc.) and do not describe the same problem.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a strong global representative for a duplicate set. If forced, issue 25725 ('Meta-Transformer') is the broadest title, but it is still a distinct request rather than a duplicate of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:19631", + "right": "issue:22372", + "accept": false, + "reason": "EDSR/MDSR and Restormer are different image restoration models; same broad domain, but not the same request." + }, + { + "left": "issue:19631", + "right": "issue:20362", + "accept": false, + "reason": "EDSR/MDSR and FlexiBERT are unrelated model requests in different areas." + }, + { + "left": "issue:19865", + "right": "issue:25725", + "accept": false, + "reason": "VATT and Meta-Transformer are different architectures with no evidence of the same code path or change." + }, + { + "left": "issue:19865", + "right": "issue:22372", + "accept": false, + "reason": "VATT and Restormer are separate model additions, not duplicates." + }, + { + "left": "issue:20334", + "right": "issue:22352", + "accept": false, + "reason": "'longformer_content' and Whisper XVector finetuning are unrelated tasks/models." + }, + { + "left": "issue:20334", + "right": "issue:20362", + "accept": false, + "reason": "Longformer-content and FlexiBERT are different model requests." + }, + { + "left": "issue:20334", + "right": "issue:25725", + "accept": false, + "reason": "Longformer-content and Meta-Transformer target different models/features." + }, + { + "left": "issue:20362", + "right": "issue:25839", + "accept": false, + "reason": "FlexiBERT and D_Nikud are distinct model additions with no shared underlying bug/change." + }, + { + "left": "issue:20362", + "right": "issue:25725", + "accept": false, + "reason": "FlexiBERT and Meta-Transformer are unrelated model requests." + }, + { + "left": "issue:22352", + "right": "issue:25725", + "accept": false, + "reason": "Whisper XVector finetuning is not the same as adding Meta-Transformer." + }, + { + "left": "issue:22372", + "right": "issue:25725", + "accept": false, + "reason": "Restormer and Meta-Transformer are different models; no duplicate evidence beyond broad 'add model' similarity." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 5, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-26350-16", + "nodes": [ + "issue:26350", + "issue:26794", + "issue:27088", + "issue:28936", + "issue:30191", + "issue:30240", + "issue:33260", + "issue:38435", + "issue:43834", + "issue:44365", + "issue:44370", + "issue:46222", + "issue:46226", + "issue:46931", + "issue:47299", + "issue:47402" + ], + "soft_pairs": [ + "issue:26350|issue:33260", + "issue:26794|issue:44370", + "issue:26794|issue:44365", + "issue:26794|issue:47402", + "issue:26794|issue:43834", + "issue:26794|issue:47299", + "issue:26794|issue:27088", + "issue:27088|issue:44370", + "issue:27088|issue:44365", + "issue:27088|issue:47402", + "issue:27088|issue:43834", + "issue:28936|issue:43834", + "issue:27088|issue:28936", + "issue:28936|issue:44365", + "issue:28936|issue:30240", + "issue:28936|issue:38435", + "issue:27088|issue:30191", + "issue:26794|issue:30191", + "issue:30191|issue:30240", + "issue:28936|issue:30191", + "issue:26794|issue:30240", + "issue:30240|issue:44370", + "issue:30240|issue:44365", + "issue:30240|issue:47402", + "issue:30240|issue:43834", + "issue:38435|issue:44370", + "issue:38435|issue:44365", + "issue:38435|issue:47402", + "issue:38435|issue:43834", + "issue:38435|issue:47299", + "issue:43834|issue:44370", + "issue:43834|issue:44365", + "issue:43834|issue:47402", + "issue:43834|issue:47299", + "issue:43834|issue:46226", + "issue:44365|issue:44370", + "issue:44365|issue:47402", + "issue:44365|issue:47299", + "issue:44365|issue:46226", + "issue:44370|issue:47402", + "issue:44370|issue:47299", + "issue:44370|issue:46226", + "issue:46222|issue:47299", + "issue:44370|issue:46222", + "issue:44365|issue:46222", + "issue:46222|issue:47402", + "issue:43834|issue:46222", + "issue:46226|issue:47402", + "issue:46226|issue:46931", + "issue:44370|issue:46931", + "issue:44365|issue:46931", + "issue:43834|issue:46931", + "issue:46931|issue:47402", + "issue:47299|issue:47402", + "issue:46226|issue:47299" + ], + "prepared_review_unit_hash": "03cc2ddfa384ca34997e8c0de08d525e48e3607602eb4f715a47e3bc935652e3", + "component_budget": { + "node_count": 16, + "item_count": 16, + "soft_pair_count": 55, + "serialized_chars": 29306, + "estimated_input_tokens": 7327, + "estimated_eval_tokens": 14910 + }, + "budget": { + "node_count": 16, + "item_count": 16, + "soft_pair_count": 55, + "serialized_chars": 29306, + "estimated_input_tokens": 7327, + "estimated_eval_tokens": 14910 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is heterogeneous and appears to contain multiple unrelated themes: two architecture-support feature requests and a large set of i18n translation template issues. The soft similarities are mostly template-driven, not evidence of true duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "If forced to choose one issue as the anchor, issue 43834 is the most central representative of the dense i18n translation subgroup, but the overall cluster is not a true duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43834 is the best single issue to represent the largest near-identical translation-template subgroup, though it should not be merged with the other language requests as duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Different feature requests: Flash Attention 2 support vs GGUF support for more architectures." + }, + { + "left": "issue:26794", + "right": "issue:44370", + "accept": false, + "reason": "Both are i18n translation requests, but they appear to be separate language tasks rather than the same underlying issue." + }, + { + "left": "issue:26794", + "right": "issue:44365", + "accept": false, + "reason": "Same translation-program template, but no evidence it targets the same language or change." + }, + { + "left": "issue:26794", + "right": "issue:47402", + "accept": false, + "reason": "Translation-template similarity only; likely a different language request." + }, + { + "left": "issue:26794", + "right": "issue:43834", + "accept": false, + "reason": "Similar i18n template, but not enough evidence that these are the same translation task." + }, + { + "left": "issue:26794", + "right": "issue:47299", + "accept": false, + "reason": "Different translation request; similar formatting is not enough to call them duplicates." + }, + { + "left": "issue:26794", + "right": "issue:27088", + "accept": false, + "reason": "Kannada vs Turkish translation requests are distinct issues." + }, + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Different language translation requests; only the template matches." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "Template overlap only; no concrete shared translation target." + }, + { + "left": "issue:27088", + "right": "issue:47402", + "accept": false, + "reason": "Different i18n request, not the same underlying change." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "Same translation template, but separate language-specific issues." + }, + { + "left": "issue:28936", + "right": "issue:43834", + "accept": false, + "reason": "Spanish translation request vs generic i18n template; not clearly the same issue." + }, + { + "left": "issue:27088", + "right": "issue:28936", + "accept": false, + "reason": "Turkish and Spanish translation issues are different language tasks." + }, + { + "left": "issue:28936", + "right": "issue:44365", + "accept": false, + "reason": "Different language request; similarity is only in the issue template." + }, + { + "left": "issue:28936", + "right": "issue:30240", + "accept": false, + "reason": "Spanish vs Polish translation requests are distinct." + }, + { + "left": "issue:28936", + "right": "issue:38435", + "accept": false, + "reason": "Different language translation issues; no evidence of duplication." + }, + { + "left": "issue:27088", + "right": "issue:30191", + "accept": false, + "reason": "Turkish translation vs Persian translation; separate issues." + }, + { + "left": "issue:26794", + "right": "issue:30191", + "accept": false, + "reason": "Kannada and Persian translation tasks are not the same change." + }, + { + "left": "issue:30191", + "right": "issue:30240", + "accept": false, + "reason": "Persian vs Polish translation requests are distinct language tasks." + }, + { + "left": "issue:28936", + "right": "issue:30191", + "accept": false, + "reason": "Spanish and Persian translation requests are separate issues." + }, + { + "left": "issue:26794", + "right": "issue:30240", + "accept": false, + "reason": "Different target languages, so not duplicates." + }, + { + "left": "issue:30240", + "right": "issue:44370", + "accept": false, + "reason": "Polish translation request vs generic placeholder translation issue; no proof they are the same." + }, + { + "left": "issue:30240", + "right": "issue:44365", + "accept": false, + "reason": "Polish vs another i18n template issue; similar structure only." + }, + { + "left": "issue:30240", + "right": "issue:47402", + "accept": false, + "reason": "Different translation task, despite near-identical templated wording." + }, + { + "left": "issue:30240", + "right": "issue:43834", + "accept": false, + "reason": "Polish translation issue and a generic i18n placeholder are not enough to merge." + }, + { + "left": "issue:38435", + "right": "issue:44370", + "accept": false, + "reason": "Romanian translation request vs another i18n template issue; likely different language." + }, + { + "left": "issue:38435", + "right": "issue:44365", + "accept": false, + "reason": "Template overlap only; distinct translation requests." + }, + { + "left": "issue:38435", + "right": "issue:47402", + "accept": false, + "reason": "Romanian translation request is separate from the generic placeholder issue." + }, + { + "left": "issue:38435", + "right": "issue:43834", + "accept": false, + "reason": "Different language tasks; not the same underlying issue." + }, + { + "left": "issue:38435", + "right": "issue:47299", + "accept": false, + "reason": "Romanian translation issue and the Thai-script issue are distinct." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": false, + "reason": "Both are templated i18n issues, but there is no evidence they refer to the same language." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": false, + "reason": "Near-identical template, but not enough to collapse separate translation requests." + }, + { + "left": "issue:43834", + "right": "issue:47402", + "accept": false, + "reason": "Same template family only; likely different language target." + }, + { + "left": "issue:43834", + "right": "issue:47299", + "accept": false, + "reason": "Different underlying translation request, despite templated similarity." + }, + { + "left": "issue:43834", + "right": "issue:46226", + "accept": false, + "reason": "The issues look templated, but no evidence they are the same language or task." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": false, + "reason": "Placeholder i18n titles only; not enough to show the same translation change." + }, + { + "left": "issue:44365", + "right": "issue:47402", + "accept": false, + "reason": "Same template, but likely different language requests." + }, + { + "left": "issue:44365", + "right": "issue:47299", + "accept": false, + "reason": "Distinct translation issue; template similarity is insufficient." + }, + { + "left": "issue:44365", + "right": "issue:46226", + "accept": false, + "reason": "No concrete sign these are the same translation task." + }, + { + "left": "issue:44370", + "right": "issue:47402", + "accept": false, + "reason": "Templated i18n issues only; insufficient evidence of duplication." + }, + { + "left": "issue:44370", + "right": "issue:47299", + "accept": false, + "reason": "Different translation request, not the same underlying issue." + }, + { + "left": "issue:44370", + "right": "issue:46226", + "accept": false, + "reason": "Same template family, but likely separate language tasks." + }, + { + "left": "issue:46222", + "right": "issue:47299", + "accept": false, + "reason": "Different-language/locale-style issues; similarity is only superficial." + }, + { + "left": "issue:44370", + "right": "issue:46222", + "accept": false, + "reason": "Templated i18n issue vs Thai-script issue; no evidence of the same request." + }, + { + "left": "issue:44365", + "right": "issue:46222", + "accept": false, + "reason": "Different content despite superficial template similarity." + }, + { + "left": "issue:46222", + "right": "issue:47402", + "accept": false, + "reason": "Separate translation requests; not the same underlying change." + }, + { + "left": "issue:43834", + "right": "issue:46222", + "accept": false, + "reason": "No indication these are the same language translation task." + }, + { + "left": "issue:46226", + "right": "issue:47402", + "accept": false, + "reason": "Different issues with only templated resemblance." + }, + { + "left": "issue:46226", + "right": "issue:46931", + "accept": false, + "reason": "Distinct Thai-language issues; not enough to merge as duplicates." + }, + { + "left": "issue:44370", + "right": "issue:46931", + "accept": false, + "reason": "Different translation/localization request; similarity is not concrete duplication." + }, + { + "left": "issue:44365", + "right": "issue:46931", + "accept": false, + "reason": "Template similarity only; likely separate topics." + }, + { + "left": "issue:43834", + "right": "issue:46931", + "accept": false, + "reason": "The titles are generically similar, but the issues appear to be different requests." + }, + { + "left": "issue:46931", + "right": "issue:47402", + "accept": false, + "reason": "Different translation issues; no evidence they are the same change." + }, + { + "left": "issue:47299", + "right": "issue:47402", + "accept": false, + "reason": "Different language/localization issues, not the same underlying request." + }, + { + "left": "issue:46226", + "right": "issue:47299", + "accept": false, + "reason": "Separate issues with only superficial textual similarity." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 6, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-30396-2", + "nodes": [ + "issue:30396", + "issue:31103" + ], + "soft_pairs": [ + "issue:30396|issue:31103" + ], + "prepared_review_unit_hash": "5ff1df7fc293c7b99c7336f46903d747fb1c290894c97d33aa3f6bb72c125b26", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1540, + "estimated_input_tokens": 385, + "estimated_eval_tokens": 1026 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1540, + "estimated_input_tokens": 385, + "estimated_eval_tokens": 1026 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are closely related HelpingAI requests, but the packet only provides titles and soft similarity. That is not enough to conclude they are the same underlying request; #31103 could be an implementation/integration task rather than the specific model-addition request in #30396.", + "confidence": 0.55, + "canonical_issue_reason": "#30396 is the clearer and more specific request title, so it is the better representative of the cluster if these are grouped at all.", + "canonical_pr_reason": null, + "best_issue_reason": "#30396 is the best issue to surface because it names the concrete HelpingAI-3B-v2.2 request more explicitly than #31103, while still avoiding overclaiming that the two are definite duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:30396", + "right": "issue:31103", + "accept": false, + "reason": "The titles are similar, but we only have soft similarity and no body/link evidence. #31103 may be an implementation or integration task related to #30396 rather than the same exact request, so this edge should stay rejected." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary goes a bit beyond the packet evidence. We only have titles and a soft-similarity signal, so saying both issues are the same underlying feature request is stronger than warranted. In particular, issue 31103 (\u201cAdd HelpingAIForCausalLM to it\u201d) could be an implementation/integration task related to, but not necessarily the same as, the specific model request in 30396. The soft-edge verdict should stay more conservative unless the bodies or links confirm the match." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 7, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-36718-23", + "nodes": [ + "issue:36718", + "issue:36817", + "issue:36879", + "issue:37015", + "issue:37334", + "issue:40330", + "issue:41115", + "issue:41211", + "issue:42503", + "issue:42738", + "issue:42954", + "issue:43671", + "issue:43837", + "issue:43909", + "issue:44121", + "issue:44205", + "issue:45027", + "issue:45177", + "issue:45322", + "issue:46133", + "issue:46645", + "issue:46653", + "issue:47004" + ], + "soft_pairs": [ + "issue:36718|issue:47004", + "issue:36718|issue:41115", + "issue:36817|issue:41211", + "issue:36817|issue:43671", + "issue:36817|issue:36879", + "issue:36817|issue:42954", + "issue:36879|issue:41211", + "issue:36879|issue:43909", + "issue:36879|issue:44121", + "issue:36879|issue:43837", + "issue:36879|issue:43671", + "issue:37015|issue:43837", + "issue:37015|issue:44121", + "issue:37015|issue:43909", + "issue:37015|issue:37334", + "issue:36879|issue:37015", + "issue:37334|issue:41115", + "issue:37334|issue:43837", + "issue:37334|issue:43671", + "issue:37334|issue:42503", + "issue:40330|issue:42503", + "issue:40330|issue:43671", + "issue:40330|issue:42954", + "issue:40330|issue:46653", + "issue:40330|issue:42738", + "issue:41115|issue:43837", + "issue:41115|issue:44121", + "issue:41115|issue:42503", + "issue:41211|issue:43671", + "issue:37015|issue:42503", + "issue:42503|issue:43837", + "issue:42738|issue:45027", + "issue:42503|issue:42738", + "issue:42738|issue:42954", + "issue:42738|issue:46653", + "issue:42954|issue:44205", + "issue:37334|issue:42954", + "issue:42954|issue:43671", + "issue:42503|issue:42954", + "issue:43671|issue:43837", + "issue:43671|issue:45027", + "issue:43671|issue:45177", + "issue:43671|issue:44121", + "issue:43837|issue:43909", + "issue:43837|issue:45322", + "issue:43837|issue:44121", + "issue:43909|issue:44205", + "issue:43909|issue:46645", + "issue:44121|issue:45322", + "issue:41115|issue:45027", + "issue:43837|issue:45027", + "issue:37015|issue:45027", + "issue:37334|issue:45027", + "issue:43837|issue:45177", + "issue:42954|issue:45177", + "issue:45322|issue:46133", + "issue:42954|issue:46653", + "issue:42503|issue:46653", + "issue:45027|issue:46653" + ], + "prepared_review_unit_hash": "5cafd11c891e9c8332d2c75ac8473c79f8421e6ef7b923b1e76a3e1c20c98db6", + "component_budget": { + "node_count": 23, + "item_count": 23, + "soft_pair_count": 59, + "serialized_chars": 33618, + "estimated_input_tokens": 8405, + "estimated_eval_tokens": 17066 + }, + "budget": { + "node_count": 23, + "item_count": 23, + "soft_pair_count": 59, + "serialized_chars": 33618, + "estimated_input_tokens": 8405, + "estimated_eval_tokens": 17066 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is heterogeneous: it mixes unrelated model requests across vision, audio, NLP, keypoint matching, and config/model-architecture updates. The soft similarities are mostly driven by generic 'add/support model' wording, not the same underlying change, so none should be deduplicated.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36718", + "right": "issue:47004", + "accept": false, + "reason": "Both are keypoint-related, but RoMa matcher vs LoMa detector+matcher are different model requests and code paths." + }, + { + "left": "issue:36718", + "right": "issue:41115", + "accept": false, + "reason": "RoMa keypoint matching and MiniCPM3 architecture are unrelated model additions." + }, + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "EuroBert config and DEIMv2 are different models with different integration work." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBert config and Qwen3-TTS are different model families and modalities." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "EuroBert config and RF-DETR model requests are unrelated." + }, + { + "left": "issue:36817", + "right": "issue:42954", + "accept": false, + "reason": "EuroBert and XiaoMi MiMo are distinct model additions." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR and DEIMv2 are different vision models, not the same change." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "RF-DETR and LFM2.5 Audio target different modalities and code paths." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "RF-DETR and OpenAI weight-sparse transformer are unrelated model requests." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "RF-DETR and Qwen3-ASR are different model families and tasks." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR vision detector request is unrelated to Qwen3-TTS audio support." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT and Qwen3-ASR are unrelated model additions." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT and circuit-sparsity/weight-sparse transformer are different architectures and requests." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT and LFM2.5 Audio are different modalities and model families." + }, + { + "left": "issue:37015", + "right": "issue:37334", + "accept": false, + "reason": "NeoBERT and LLMDet are distinct models; not the same underlying change." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR and NeoBERT are unrelated model additions." + }, + { + "left": "issue:37334", + "right": "issue:41115", + "accept": false, + "reason": "LLMDet support and MiniCPM3 architecture are different integration efforts." + }, + { + "left": "issue:37334", + "right": "issue:43837", + "accept": false, + "reason": "LLMDet and Qwen3-ASR are different model requests." + }, + { + "left": "issue:37334", + "right": "issue:43671", + "accept": false, + "reason": "LLMDet and Qwen3-TTS are unrelated model families and tasks." + }, + { + "left": "issue:37334", + "right": "issue:42503", + "accept": false, + "reason": "LLMDet and ModernVBERT are different model additions." + }, + { + "left": "issue:40330", + "right": "issue:42503", + "accept": false, + "reason": "open-qwen2vl-base and ModernVBERT are different models despite both being vision-language oriented." + }, + { + "left": "issue:40330", + "right": "issue:43671", + "accept": false, + "reason": "open-qwen2vl-base and Qwen3-TTS are different tasks and implementations." + }, + { + "left": "issue:40330", + "right": "issue:42954", + "accept": false, + "reason": "open-qwen2vl-base and XiaoMi MiMo are unrelated requests." + }, + { + "left": "issue:40330", + "right": "issue:46653", + "accept": false, + "reason": "open-qwen2vl-base and Mirko Privitera are clearly unrelated." + }, + { + "left": "issue:40330", + "right": "issue:42738", + "accept": false, + "reason": "open-qwen2vl-base and BERT-like models with RoPE are not the same change." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 architecture and Qwen3-ASR support are unrelated model requests." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 and the weight-sparse transformer request target different model families." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 architecture and ModernVBERT models are different additions." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are unrelated models and modalities." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT are distinct BERT-like model requests, not the same underlying change." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT and Qwen3-ASR are different model families and tasks." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like RoPE models and Voxtral TTS support are unrelated." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT and generic BERT-like RoPE support overlap in theme but not in concrete requested model or code path." + }, + { + "left": "issue:42738", + "right": "issue:42954", + "accept": false, + "reason": "RoPE BERT-like models and XiaoMi MiMo are different requests." + }, + { + "left": "issue:42738", + "right": "issue:46653", + "accept": false, + "reason": "RoPE BERT-like models and Mirko Privitera are unrelated." + }, + { + "left": "issue:42954", + "right": "issue:44205", + "accept": false, + "reason": "XiaoMi MiMo and SAM3-LiteText are different model additions." + }, + { + "left": "issue:37334", + "right": "issue:42954", + "accept": false, + "reason": "LLMDet and XiaoMi MiMo are unrelated model requests." + }, + { + "left": "issue:42954", + "right": "issue:43671", + "accept": false, + "reason": "XiaoMi MiMo and Qwen3-TTS are different models and modalities." + }, + { + "left": "issue:42503", + "right": "issue:42954", + "accept": false, + "reason": "ModernVBERT and XiaoMi MiMo are unrelated additions." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS and Qwen3-ASR are sibling tasks but different concrete model support changes." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS and Voxtral TTS support are different TTS model requests, not the same change." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS and DeepSeek-OCR2 are different modalities and code paths." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS and the weight-sparse transformer request are unrelated." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR and LFM2.5 Audio are different audio model requests." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR and EUPE are different model families." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-ASR and the weight-sparse transformer request are unrelated." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "LFM2.5 Audio and SAM3-LiteText are different modalities and integrations." + }, + { + "left": "issue:43909", + "right": "issue:46645", + "accept": false, + "reason": "LFM2.5 Audio and Granite SWA+sink option are unrelated changes." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "OpenAI weight-sparse transformer and EUPE are different model requests." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 and Voxtral TTS support are unrelated." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS support are different audio tasks and models." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS support are unrelated." + }, + { + "left": "issue:37334", + "right": "issue:45027", + "accept": false, + "reason": "LLMDet and Voxtral TTS support are unrelated." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model requests." + }, + { + "left": "issue:42954", + "right": "issue:45177", + "accept": false, + "reason": "XiaoMi MiMo and DeepSeek-OCR2 are unrelated." + }, + { + "left": "issue:45322", + "right": "issue:46133", + "accept": false, + "reason": "EUPE and TIPSv2 are both research-model requests but they are different architectures and not the same change." + }, + { + "left": "issue:42954", + "right": "issue:46653", + "accept": false, + "reason": "XiaoMi MiMo and Mirko Privitera are unrelated." + }, + { + "left": "issue:42503", + "right": "issue:46653", + "accept": false, + "reason": "ModernVBERT and Mirko Privitera are unrelated." + }, + { + "left": "issue:45027", + "right": "issue:46653", + "accept": false, + "reason": "Voxtral TTS support and Mirko Privitera are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 8, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-38740-2", + "nodes": [ + "issue:38740", + "issue:46443" + ], + "soft_pairs": [ + "issue:38740|issue:46443" + ], + "prepared_review_unit_hash": "ff253ac42ef146eecdef940b866acab7b560f5c710b0b89b8bc205c74141006b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1505, + "estimated_input_tokens": 377, + "estimated_eval_tokens": 1010 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1505, + "estimated_input_tokens": 377, + "estimated_eval_tokens": 1010 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are not duplicates: one is a documentation request to add Pruna as an optimization framework, while the other is a technical Vision RoPE refactor. They describe different domains and different kinds of work.", + "confidence": 0.98, + "canonical_issue_reason": "Neither issue is a clear canonical duplicate of the other; they are unrelated topics.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #46443 is the better standalone representative because it is a concrete technical refactor, whereas #38740 is only a docs request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38740", + "right": "issue:46443", + "accept": false, + "reason": "Different subject matter: documentation/integration request versus Vision RoPE refactor. No shared code path or duplicate bug/change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 9, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41084-2", + "nodes": [ + "issue:41084", + "issue:43037" + ], + "soft_pairs": [ + "issue:41084|issue:43037" + ], + "prepared_review_unit_hash": "805f5e6ae6c5faac6369a1ab9f78c2ba41f4c3adf5f109176fe01434bb549a60", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1479, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1479, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates. \"Set Block Decoding\" is about a decoding behavior/code-path change, while \"DeepSeek v3.2 support\" is a model-specific support request. The soft similarity is superficial only.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different underlying requests with no clear shared bug or change: block decoding behavior vs support for a specific model version." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 10, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41306-2", + "nodes": [ + "issue:41306", + "issue:41826" + ], + "soft_pairs": [ + "issue:41306|issue:41826" + ], + "prepared_review_unit_hash": "7d815a7c8c225ccd997c83281b525450975b799b60552b007900e98e34e26afd", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are separate enhancement issues: one requests a `from_pretrained()` flag for TiledFusedLogitsLoss/FLCE, while the other asks for integrating TiledMLP to reduce memory footprint. They target different components and underlying changes.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 41306 is the more concrete, narrowly scoped request about exposing a specific `from_pretrained()` flag for FLCE/TiledFusedLogitsLoss.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 41306 is the better representative because it has a clear API change request and is more self-contained than the broader TiledMLP integration idea.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41306", + "right": "issue:41826", + "accept": false, + "reason": "Different underlying changes: one is an API flag for loading a specific loss module, the other is a separate memory-optimization integration for TiledMLP. They are not the same bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43598-2", + "nodes": [ + "issue:43598", + "issue:43600" + ], + "soft_pairs": [ + "issue:43598|issue:43600" + ], + "prepared_review_unit_hash": "2f974ed6de37fd502ef86a0974cd99d68f65403edcd22ff7e38d07dd8d9aa2a0", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both issues are about Trainer extensibility, but they describe different bugs: one is about `remove_unused_column` behavior, the other about FLOPS estimation for custom trainers. They are related, not duplicates.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 43598 is the better representative because it targets a concrete, user-facing Trainer behavior change (`remove_unused_column`) with a clear code path to adjust.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43598 is slightly stronger as the canonical issue because it is more specific and actionable; 43600 is a separate accounting/metrics problem.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "Different underlying problems: one concerns automatic column removal in Trainer, the other concerns FLOPS counting for custom trainers. They share the Trainer/customization theme but are not the same fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 12, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43684-2", + "nodes": [ + "issue:43684", + "issue:44405" + ], + "soft_pairs": [ + "issue:43684|issue:44405" + ], + "prepared_review_unit_hash": "185c61c718d6759947d1a5ae4055f36cffe5bbb86191d90cd467c15145281425", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related only at a very high level: both ask for additional AutoModel support for Qwen variants, but they target different model families and different integration points. They should not be treated as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 43684 is the broader request about registering Qwen3-Omni with AutoModel/AutoModelForConditionalGeneration, so it is the better representative of the general auto-model registration theme.", + "canonical_pr_reason": null, + "best_issue_reason": "43684 is the most general and central issue in this pair; 44405 is a separate, more specific request for AutoModelForSequenceClassification on Qwen3.5.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different underlying change requests: Qwen3-Omni auto-registration for generation vs Qwen3.5 sequence-classification support. Same broad subsystem, but not the same bug or mergeable fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43835-4", + "nodes": [ + "issue:43835", + "issue:43845", + "issue:43846", + "issue:43859" + ], + "soft_pairs": [ + "issue:43835|issue:43845", + "issue:43835|issue:43846", + "issue:43835|issue:43859", + "issue:43845|issue:43846", + "issue:43845|issue:43859", + "issue:43846|issue:43859" + ], + "prepared_review_unit_hash": "579ff5b772d133b7198f701664453049b4c2fb926cd0bf1491b931b6e28f11bd", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4264, + "estimated_input_tokens": 1066, + "estimated_eval_tokens": 2388 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4264, + "estimated_input_tokens": 1066, + "estimated_eval_tokens": 2388 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are extremely sparse issue reports: one unrelated title ('Fett') and three near-identical 'huggingface ' issues with no actionable body detail. There is not enough evidence to confidently confirm true duplicates beyond superficial title similarity.", + "confidence": 0.34, + "canonical_issue_reason": "issue:43845 is the earliest of the repeated 'huggingface ' reports and the cleanest anchor for the cluster, but the report text is too thin to establish more than superficial similarity.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43845 is the best representative issue because it is an early instance of the repeated title pattern; however, all of the issues are too under-described to be strong canonical candidates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different titles ('Fett' vs 'huggingface ') and no body detail or shared targets to show the same underlying bug." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different titles and no substantive content linking them to the same problem." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different titles and no actionable description to justify a duplicate relationship." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": false, + "reason": "The titles are the same, but both reports are too vague to confirm the same concrete issue with confidence." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": false, + "reason": "Only superficial title similarity is present; the bodies are too sparse to prove the same bug." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": false, + "reason": "Only repeated generic title text is shared; there is no evidence of the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 14, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44371-6", + "nodes": [ + "issue:44371", + "issue:44418", + "issue:44572", + "issue:44573", + "issue:44937", + "issue:47036" + ], + "soft_pairs": [ + "issue:44371|issue:44572", + "issue:44371|issue:44573", + "issue:44371|issue:44418", + "issue:44371|issue:47036", + "issue:44371|issue:44937", + "issue:44418|issue:44572", + "issue:44418|issue:44573", + "issue:44572|issue:44573", + "issue:44572|issue:47036", + "issue:44572|issue:44937", + "issue:44573|issue:47036", + "issue:44573|issue:44937" + ], + "prepared_review_unit_hash": "2cb9c05df8cf46a86091e92c74ac9b80e9f4d9448f1511c77a578075d0e26401", + "component_budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 12, + "serialized_chars": 7355, + "estimated_input_tokens": 1839, + "estimated_eval_tokens": 3934 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 12, + "serialized_chars": 7355, + "estimated_input_tokens": 1839, + "estimated_eval_tokens": 3934 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are mostly spam/generic placeholder issues with no concrete shared bug or change. The pairwise similarity is superficial and does not justify deduplication; no soft edges should be merged.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:44418 is the least-noisy and most actionable-looking item, but it is still too vague to serve as a real canonical issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both are spam-like placeholders with no describable underlying problem; no concrete duplicate signal." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Same generic spam pattern; no evidence they report the same issue." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is spam, the other is a vague documentation suggestion; not the same underlying bug/change." + }, + { + "left": "issue:44371", + "right": "issue:47036", + "accept": false, + "reason": "Spam placeholder vs. vague community tip; no shared concrete problem." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder vs. an unrelated titled issue; similarity is too weak for a duplicate." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Generic documentation request and spam placeholder are not the same change." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "No concrete overlap beyond being low-information issues." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Both are spam-like placeholders, but there is no evidence of a shared underlying bug or requested change." + }, + { + "left": "issue:44572", + "right": "issue:47036", + "accept": false, + "reason": "Spam placeholder vs. community benchmark tip; too vague to merge." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "No concrete shared issue or code path is evident." + }, + { + "left": "issue:44573", + "right": "issue:47036", + "accept": false, + "reason": "Generic placeholder and benchmark tip are not duplicates." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "No substantive evidence of the same underlying problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 15, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45145-2", + "nodes": [ + "issue:45145", + "issue:45151" + ], + "soft_pairs": [ + "issue:45145|issue:45151" + ], + "prepared_review_unit_hash": "f08750887a30026e629dd9363cb31605d6cebf6f3797312257b92a9cff0998c1", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These closed issues look like likely duplicate / very similar reports: they share the same title, have similar body lengths, and there is no filenames/targets metadata to distinguish them. #45145 is the best canonical issue because it was opened earlier and already has an inbound reference.", + "confidence": 0.86, + "canonical_issue_reason": "Issue #45145 is the better canonical issue because it was opened slightly earlier and already has an inbound reference, making it the most natural anchor for a likely-duplicate issue pair.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #45145 is the best issue representative because it is the earliest report and already has an inbound reference; the other issue appears to be a very similar duplicate report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": true, + "reason": "Same title and very similar report size, with no differing filenames or explicit targets; this supports treating them as likely duplicate / very similar reports, though not as proven identical." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The packet supports a likely duplicate relationship, but the writeup is too strong in places. We only have same title, similar body length, no filenames/targets, and a soft-similarity signal; we do not have direct evidence that the bodies are \u201ceffectively identical\u201d or that the content is \u201chighly overlapping.\u201d Please downgrade the summary and soft-edge language to something more conservative like \u201clikely duplicate / very similar reports.\u201d" + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 16, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45295-2", + "nodes": [ + "issue:45295", + "issue:45373" + ], + "soft_pairs": [], + "prepared_review_unit_hash": "f5fd95fada2f530ab0999e343c17c16266cde0c7182bb22509972ca44827cd49", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 0, + "serialized_chars": 1286, + "estimated_input_tokens": 322, + "estimated_eval_tokens": 900 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 0, + "serialized_chars": 1286, + "estimated_input_tokens": 322, + "estimated_eval_tokens": 900 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues describe the same missing Gemma 4 sequence-classification support; the later, more specific open issue is the better cluster representative.", + "confidence": 0.94, + "canonical_issue_reason": "issue #45373 is the clearest canonical tracker: it is open, explicitly names the missing class `Gemma4ForSequenceClassification`, and states the module-level gap relative to Gemma 2/3.", + "canonical_pr_reason": null, + "best_issue_reason": "issue #45373 best captures the concrete request and likely remaining work, while #45295 is a shorter earlier duplicate with less detail and is already closed.", + "best_pr_reason": null, + "soft_edge_verdicts": [] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 17, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45762-2", + "nodes": [ + "issue:45762", + "issue:45763" + ], + "soft_pairs": [ + "issue:45762|issue:45763" + ], + "prepared_review_unit_hash": "10aaabfe4079087e968315c4156f28829abc0088e84080201515fd6e449b0849", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two closed issues with unrelated person-name titles and no body or linkage; they do not look like duplicates of the same underlying bug or request.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a meaningful canonical representative here because the two issues are unrelated and appear to be non-technical spam/placeholder entries.", + "best_pr_reason": "No PRs are present in this cluster.", + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "Different names and dates, no shared context, body, or explicit linkage; not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 18, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45764-2", + "nodes": [ + "issue:45764", + "issue:45765" + ], + "soft_pairs": [ + "issue:45764|issue:45765" + ], + "prepared_review_unit_hash": "0b7efa2185c5fbfe0024b6a94a19276470f6603c78b6a6abbc8798ae65b5ed1e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two closed issues look like unrelated spam-like entries: each is just a different personal name plus a date, with no bug-report context. They should not be treated as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items have different names/dates and no shared problem description.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong best representative; if forced, 45764 is only marginally earlier, but neither issue is a meaningful cluster anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45764", + "right": "issue:45765", + "accept": false, + "reason": "Reject: different personal-name/date strings with no evidence of the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 19, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45810-2", + "nodes": [ + "issue:45810", + "issue:46295" + ], + "soft_pairs": [ + "issue:45810|issue:46295" + ], + "prepared_review_unit_hash": "159856e189840e7da83d431707305e047e4210335189f2735c48a10d8698071d", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are unrelated: one requests a new Qwen3_5 token-classification model class, while the other asks for FP8 training support. They should not be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46295 is the stronger standalone tracker in this pair because it has broader scope and more inbound attention, but it is not a duplicate of 45810.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45810", + "right": "issue:46295", + "accept": false, + "reason": "Different underlying requests and code paths: model-class support for token classification vs training precision/FP8 support." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 20, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45869-5", + "nodes": [ + "issue:45869", + "issue:45870", + "issue:45871", + "issue:45872", + "issue:45873" + ], + "soft_pairs": [ + "issue:45869|issue:45870", + "issue:45869|issue:45871", + "issue:45870|issue:45871", + "issue:45870|issue:45872", + "issue:45870|issue:45873", + "issue:45871|issue:45872", + "issue:45871|issue:45873", + "issue:45872|issue:45873" + ], + "prepared_review_unit_hash": "8e013e0a5b70691c570b48ca71e33f00ae88ff6b547342bffef112dbcfd60e42", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5726, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5726, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster contains two exact duplicate pairs and three distinct financial/DataForge-related issues. Accept the two exact-duplicate pairs (45870/45871 and 45872/45873); reject the cross-pair similarity matches and the dataset-vs-example match.", + "confidence": 0.96, + "canonical_issue_reason": "issue:45870 is the best canonical issue for its duplicate pair because 45871 is identical, and 45870 is the earlier report with a clear, specific title.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45870 is the strongest issue representative in this cluster because it is an exact duplicate target with the cleanest, earliest metadata; the other issues are thematically related but not the same underlying request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45869", + "right": "issue:45870", + "accept": false, + "reason": "Different artifacts: a dataset bundle offer vs a DataForge prediction example. Same financial theme, but not the same issue." + }, + { + "left": "issue:45869", + "right": "issue:45871", + "accept": false, + "reason": "Different artifacts: a dataset bundle offer vs a DataForge prediction example. Shared topic is too broad to treat as duplicates." + }, + { + "left": "issue:45870", + "right": "issue:45871", + "accept": true, + "reason": "Exact duplicate pair: identical title and matching metadata/body length, so they appear to describe the same underlying issue." + }, + { + "left": "issue:45870", + "right": "issue:45872", + "accept": false, + "reason": "Related DataForge financial content, but one is a prediction example and the other is a scraping-service request; not the same underlying issue." + }, + { + "left": "issue:45870", + "right": "issue:45873", + "accept": false, + "reason": "Related DataForge financial content, but one is a prediction example and the other is a scraping-service request; not the same underlying issue." + }, + { + "left": "issue:45871", + "right": "issue:45872", + "accept": false, + "reason": "Related financial/DataForge topic, but these describe different concrete requests: prediction example vs scraping service." + }, + { + "left": "issue:45871", + "right": "issue:45873", + "accept": false, + "reason": "Related financial/DataForge topic, but these describe different concrete requests: prediction example vs scraping service." + }, + { + "left": "issue:45872", + "right": "issue:45873", + "accept": true, + "reason": "Exact duplicate pair: identical title and matching metadata/body length, so they appear to describe the same underlying issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is grounded in the packet: 45870/45871 and 45872/45873 are the only pairs with identical titles and equal body lengths, while the remaining cross-pair matches are only broad soft similarities and are conservatively rejected. The canonical choice of the earlier report in each duplicate pair is also reasonable." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 21, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46439-2", + "nodes": [ + "issue:46439", + "issue:46441" + ], + "soft_pairs": [ + "issue:46439|issue:46441" + ], + "prepared_review_unit_hash": "54a71e62c711c52c0a1ef241f1d7049bc492260affb82df3347e253e3f45ec33", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1641, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1641, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both issues involve `cache_implementation=\"static\"` and linear-attention/Qwen3.5 failures, but they report different concrete breakages: one is an `early_initialization` shape corruption leading to `RuntimeError`, the other is a missing `linear_attention` cache path causing `KeyError`. Related area, but not the same underlying bug.", + "confidence": 0.78, + "canonical_issue_reason": "Issue 46439 is the better canonical issue because it names the more specific mechanism (`early_initialization`) and the observed failure mode, making it a stronger anchor for the static-cache/linear-attention bug family.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46439 is the best representative of the cluster because it describes the broader root-cause-style failure, while 46441 looks more like a specific downstream symptom for a particular model/configuration.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46439", + "right": "issue:46441", + "accept": false, + "reason": "Do not merge: both touch static cache and linear attention, but the reported problems differ materially (`RuntimeError` from shape corruption vs `KeyError` for missing `linear_attention`). They are related symptoms, not clearly the same defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 22, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46551-2", + "nodes": [ + "issue:46551", + "issue:46552" + ], + "soft_pairs": [ + "issue:46551|issue:46552" + ], + "prepared_review_unit_hash": "90927896cee9f5f7dc604b7b3b5cec7fb0dd1a8150dc5d519a23a627d018c754", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1709, + "estimated_input_tokens": 428, + "estimated_eval_tokens": 1112 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1709, + "estimated_input_tokens": 428, + "estimated_eval_tokens": 1112 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two closed issues appear to be duplicates: they have identical titles, the same body length, and strong soft-similarity around the same Wav2Vec2CTCTokenizer v5 bug affecting get_special_tokens_mask and convert_ids_to_tokens. The packet does not prove the body text is exactly identical, but it is still strong duplicate evidence.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #46552 is the better canonical target because it reports the same bug as #46551 and has slightly higher discussion/inbound reference activity, making it the more central duplicate thread.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #46552 is the best representative for the cluster because it fully captures the underlying tokenizer bug and has the stronger conversation/reference signal.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46551", + "right": "issue:46552", + "accept": true, + "reason": "Accept: identical titles, equal body length, and explicit soft-similarity indicate the same underlying issue report, even though exact body text identity is not proven." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The duplicate conclusion is supported, but the summary overstates the evidence by saying the issues have identical bodies. The packet directly shows identical titles, equal body length, and strong soft-similarity, but not the exact body text. The canonical/best issue choice is otherwise reasonable given the higher discussion and inbound reference activity on #46552." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 23, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46657-2", + "nodes": [ + "issue:46657", + "issue:46692" + ], + "soft_pairs": [ + "issue:46657|issue:46692" + ], + "prepared_review_unit_hash": "5ad904b4ded6bb59f98ffb382a2902d3dee2f20ffdd08438530ae47e673b0ad3", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1570, + "estimated_input_tokens": 393, + "estimated_eval_tokens": 1042 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1570, + "estimated_input_tokens": 393, + "estimated_eval_tokens": 1042 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are only loosely related by being training-time problems in detection models, but they describe different failure modes: a tensor size mismatch versus an out-of-memory condition after several epochs. They do not look like duplicates.", + "confidence": 0.95, + "canonical_issue_reason": "Issue #46657 is the better canonical issue because it reports a concrete, specific runtime error during object detection training, making it the clearer representative bug report.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #46657 is the best issue overall because it is more specific and actionable; #46692 appears to be a different symptom and likely a different underlying problem.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46657", + "right": "issue:46692", + "accept": false, + "reason": "Reject: the reported failures are different in kind and likely in root cause. #46657 is a tensor shape/size mismatch, while #46692 is an out-of-memory issue after some epochs during RT-DETR v2 training. Same broad training area, but not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 24, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46822-2", + "nodes": [ + "issue:46822", + "issue:46823" + ], + "soft_pairs": [ + "issue:46822|issue:46823" + ], + "prepared_review_unit_hash": "363c1012dd0fe4dff99ea18c9847c533e833f7660c7272e19152d112f3343f37", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two separate model-loading bugs: one about ZeRO-3 zero.Init failing to partition a composite multimodal submodule and causing OOM, and one about 4-bit auto device mapping sending a model to CPU despite sufficient memory. They do not look like duplicates or the same underlying fix.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 46822 is the more concrete, unresolved bug report with a specific ZeRO-3 partitioning failure and clear OOM symptom.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46822 is the best representative issue in this cluster because it has a precise failure mode and is still open.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46822", + "right": "issue:46823", + "accept": false, + "reason": "Different subsystems and failure modes: ZeRO-3 partitioning/OOM vs 4-bit device_map auto-placement/CPU dispatch ValueError. They are not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 25, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46906-23", + "nodes": [ + "issue:46906", + "issue:46940", + "issue:46948", + "issue:46969", + "issue:46985", + "issue:46998", + "issue:47037", + "issue:47049", + "issue:47053", + "issue:47091", + "issue:47102", + "issue:47120", + "issue:47182", + "issue:47227", + "issue:47256", + "issue:47265", + "issue:47279", + "issue:47309", + "issue:47377", + "issue:47390", + "issue:47406", + "issue:47419", + "issue:47442" + ], + "soft_pairs": [ + "issue:46906|issue:46940", + "issue:46906|issue:46948", + "issue:46906|issue:46969", + "issue:46906|issue:46985", + "issue:46906|issue:47037", + "issue:46940|issue:46948", + "issue:46948|issue:46969", + "issue:46948|issue:46985", + "issue:46940|issue:46969", + "issue:46940|issue:46985", + "issue:47037|issue:47049", + "issue:46948|issue:46998", + "issue:47049|issue:47053", + "issue:46998|issue:47049", + "issue:46969|issue:46985", + "issue:46940|issue:46998", + "issue:46985|issue:47049", + "issue:47049|issue:47091", + "issue:46998|issue:47037", + "issue:46985|issue:47037", + "issue:46969|issue:47037", + "issue:47037|issue:47091", + "issue:47037|issue:47053", + "issue:46969|issue:47091", + "issue:47053|issue:47091", + "issue:47091|issue:47102", + "issue:46998|issue:47053", + "issue:47053|issue:47102", + "issue:46985|issue:46998", + "issue:46998|issue:47091", + "issue:47037|issue:47102", + "issue:46985|issue:47091", + "issue:46985|issue:47102", + "issue:47049|issue:47102", + "issue:47265|issue:47279", + "issue:47265|issue:47309", + "issue:47279|issue:47309", + "issue:47265|issue:47377", + "issue:47279|issue:47377", + "issue:47256|issue:47265", + "issue:47256|issue:47279", + "issue:47182|issue:47256", + "issue:47120|issue:47256", + "issue:47309|issue:47377", + "issue:47102|issue:47256", + "issue:47182|issue:47265", + "issue:47120|issue:47182", + "issue:47120|issue:47279", + "issue:47120|issue:47265", + "issue:47120|issue:47377", + "issue:47182|issue:47377", + "issue:47182|issue:47279", + "issue:47227|issue:47377", + "issue:47227|issue:47256", + "issue:47182|issue:47309", + "issue:47309|issue:47419", + "issue:47227|issue:47279", + "issue:47182|issue:47227", + "issue:47227|issue:47265", + "issue:47419|issue:47442", + "issue:47377|issue:47419", + "issue:47377|issue:47442", + "issue:47279|issue:47419", + "issue:47406|issue:47419", + "issue:47279|issue:47442", + "issue:47227|issue:47442", + "issue:47390|issue:47419", + "issue:47390|issue:47406", + "issue:47390|issue:47442", + "issue:47406|issue:47442", + "issue:47377|issue:47406", + "issue:47279|issue:47406" + ], + "prepared_review_unit_hash": "9c801eb099da9facd78e6b528c9faf210229be344cd8e81e82ef3d9a80c41e0b", + "component_budget": { + "node_count": 23, + "item_count": 23, + "soft_pair_count": 74, + "serialized_chars": 39466, + "estimated_input_tokens": 9867, + "estimated_eval_tokens": 19990 + }, + "budget": { + "node_count": 23, + "item_count": 23, + "soft_pair_count": 72, + "serialized_chars": 34132, + "estimated_input_tokens": 8533, + "estimated_eval_tokens": 17322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a date-stamped series of [serge] integration failure triage issues. They look like recurring daily reports, not duplicate incidents, so the soft similarity links should not be merged.", + "confidence": 0.92, + "canonical_issue_reason": "If a single representative is needed, issue 47442 is the strongest anchor: it is the most recent open triage report and has strong inbound reference activity. But the cluster itself is a recurring series, not one duplicate bug.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 47442 is the best overall representative because it is current, open, and appears to be the active landing point for later references.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46906", + "right": "issue:46940", + "accept": false, + "reason": "Same template and subsystem, but different dates imply separate triage reports rather than the same underlying failure." + }, + { + "left": "issue:46906", + "right": "issue:46948", + "accept": false, + "reason": "These are adjacent daily triage issues, not evidence of one identical bug instance." + }, + { + "left": "issue:46906", + "right": "issue:46969", + "accept": false, + "reason": "Different report dates; this looks like recurring monitoring output, not a duplicate issue." + }, + { + "left": "issue:46906", + "right": "issue:46985", + "accept": false, + "reason": "Same recurring triage series, but no sign of the same concrete incident." + }, + { + "left": "issue:46906", + "right": "issue:47037", + "accept": false, + "reason": "Separated by several days and likely different failures; too weak for duplicate merge." + }, + { + "left": "issue:46940", + "right": "issue:46948", + "accept": false, + "reason": "Highly similar titles, but they are date-specific triage entries for different days." + }, + { + "left": "issue:46948", + "right": "issue:46969", + "accept": false, + "reason": "Same recurring triage pattern, but not the same underlying bug report." + }, + { + "left": "issue:46948", + "right": "issue:46985", + "accept": false, + "reason": "Different dates and likely different integration runs; not a duplicate." + }, + { + "left": "issue:46940", + "right": "issue:46969", + "accept": false, + "reason": "Recurring report series only; no concrete shared incident identified." + }, + { + "left": "issue:46940", + "right": "issue:46985", + "accept": false, + "reason": "Same title family, but different day-specific triage issues." + }, + { + "left": "issue:47037", + "right": "issue:47049", + "accept": false, + "reason": "Adjacent daily reports, but still separate triage artifacts." + }, + { + "left": "issue:46948", + "right": "issue:46998", + "accept": false, + "reason": "The similarity is template-level; the dated reports are distinct." + }, + { + "left": "issue:47049", + "right": "issue:47053", + "accept": false, + "reason": "Same recurring triage series, but no evidence they describe one identical failure." + }, + { + "left": "issue:46998", + "right": "issue:47049", + "accept": false, + "reason": "Different dates and separate integration triage entries." + }, + { + "left": "issue:46969", + "right": "issue:46985", + "accept": false, + "reason": "Date-stamped daily triage reports; not the same bug." + }, + { + "left": "issue:46940", + "right": "issue:46998", + "accept": false, + "reason": "Same broad topic, but these are separate daily triage issues." + }, + { + "left": "issue:46985", + "right": "issue:47049", + "accept": false, + "reason": "Recurring triage template only; no shared concrete failure shown." + }, + { + "left": "issue:47049", + "right": "issue:47091", + "accept": false, + "reason": "Similar report family, but different incident dates." + }, + { + "left": "issue:46998", + "right": "issue:47037", + "accept": false, + "reason": "Daily triage issues should not be merged on title similarity alone." + }, + { + "left": "issue:46985", + "right": "issue:47037", + "accept": false, + "reason": "Different dates and likely different integration failures." + }, + { + "left": "issue:46969", + "right": "issue:47037", + "accept": false, + "reason": "Looks like the same recurring report template, not a duplicate bug." + }, + { + "left": "issue:47037", + "right": "issue:47091", + "accept": false, + "reason": "Separate daily triage artifacts, not one underlying issue." + }, + { + "left": "issue:47037", + "right": "issue:47053", + "accept": false, + "reason": "Same recurring integration triage series; no concrete duplicate signal." + }, + { + "left": "issue:46969", + "right": "issue:47091", + "accept": false, + "reason": "Different report dates; insufficient evidence of one merged incident." + }, + { + "left": "issue:47053", + "right": "issue:47091", + "accept": false, + "reason": "Similar subsystem and title, but not the same dated report." + }, + { + "left": "issue:47091", + "right": "issue:47102", + "accept": false, + "reason": "Adjacent daily triage issues, but still distinct reports." + }, + { + "left": "issue:46998", + "right": "issue:47053", + "accept": false, + "reason": "Template similarity only; these are separate daily triage entries." + }, + { + "left": "issue:47053", + "right": "issue:47102", + "accept": false, + "reason": "Different days in the same triage series, not the same bug." + }, + { + "left": "issue:46985", + "right": "issue:46998", + "accept": false, + "reason": "Same recurring report pattern, but not a duplicate incident." + }, + { + "left": "issue:46998", + "right": "issue:47091", + "accept": false, + "reason": "No evidence they refer to one specific failure; just repeated triage." + }, + { + "left": "issue:46985", + "right": "issue:47091", + "accept": false, + "reason": "Date-specific reports from the same series should remain separate." + }, + { + "left": "issue:46985", + "right": "issue:47102", + "accept": false, + "reason": "Recurring triage issues, but not the same underlying problem." + }, + { + "left": "issue:47049", + "right": "issue:47102", + "accept": false, + "reason": "Similar text, different daily report context." + }, + { + "left": "issue:47265", + "right": "issue:47279", + "accept": false, + "reason": "These are sequential daily triage issues, not one duplicated bug report." + }, + { + "left": "issue:47265", + "right": "issue:47309", + "accept": false, + "reason": "Same report template, but distinct dated incidents." + }, + { + "left": "issue:47279", + "right": "issue:47309", + "accept": false, + "reason": "High text similarity is just the daily triage format; do not merge." + }, + { + "left": "issue:47265", + "right": "issue:47377", + "accept": false, + "reason": "Different dates and likely different failure snapshots." + }, + { + "left": "issue:47279", + "right": "issue:47377", + "accept": false, + "reason": "Same recurring integration triage series, but not the same concrete issue." + }, + { + "left": "issue:47256", + "right": "issue:47265", + "accept": false, + "reason": "Adjacent daily reports, but no shared underlying bug is shown." + }, + { + "left": "issue:47256", + "right": "issue:47279", + "accept": false, + "reason": "These are separate day-by-day triage artifacts." + }, + { + "left": "issue:47182", + "right": "issue:47256", + "accept": false, + "reason": "Same family of reports, but different dates and likely different incidents." + }, + { + "left": "issue:47120", + "right": "issue:47256", + "accept": false, + "reason": "Recurring triage entry, not a duplicate fix target." + }, + { + "left": "issue:47102", + "right": "issue:47256", + "accept": false, + "reason": "Too far apart in date and context to be the same underlying bug." + }, + { + "left": "issue:47182", + "right": "issue:47265", + "accept": false, + "reason": "Both are triage reports, but separate daily instances." + }, + { + "left": "issue:47120", + "right": "issue:47182", + "accept": false, + "reason": "The similarity is only the shared reporting template." + }, + { + "left": "issue:47120", + "right": "issue:47279", + "accept": false, + "reason": "Different days in a recurring triage series." + }, + { + "left": "issue:47120", + "right": "issue:47265", + "accept": false, + "reason": "Same subsystem and title, but not the same incident." + }, + { + "left": "issue:47120", + "right": "issue:47377", + "accept": false, + "reason": "Recurring daily triage issue, not a duplicate bug." + }, + { + "left": "issue:47182", + "right": "issue:47377", + "accept": false, + "reason": "Date-stamped reports should remain distinct unless they name the same failure." + }, + { + "left": "issue:47182", + "right": "issue:47279", + "accept": false, + "reason": "Same report pattern, but separate triage entries." + }, + { + "left": "issue:47227", + "right": "issue:47377", + "accept": false, + "reason": "No evidence these point to one concrete code-path problem." + }, + { + "left": "issue:47227", + "right": "issue:47256", + "accept": false, + "reason": "Adjacent daily reports, but not duplicates." + }, + { + "left": "issue:47182", + "right": "issue:47309", + "accept": false, + "reason": "Same recurring integration triage format; different dates." + }, + { + "left": "issue:47309", + "right": "issue:47377", + "accept": false, + "reason": "These are sequential daily reports, not one merged failure." + }, + { + "left": "issue:47182", + "right": "issue:47227", + "accept": false, + "reason": "Only title/template similarity; different daily incidents." + }, + { + "left": "issue:47227", + "right": "issue:47279", + "accept": false, + "reason": "Recurring triage series, but not the same bug." + }, + { + "left": "issue:47227", + "right": "issue:47265", + "accept": false, + "reason": "Date-specific reports with no shared concrete failure evidence." + }, + { + "left": "issue:47419", + "right": "issue:47442", + "accept": false, + "reason": "Same series, but the open report is a later day and should not be merged as a duplicate." + }, + { + "left": "issue:47377", + "right": "issue:47419", + "accept": false, + "reason": "Recurring daily triage issues, not the same underlying incident." + }, + { + "left": "issue:47377", + "right": "issue:47442", + "accept": false, + "reason": "Different report dates; title similarity alone is insufficient." + }, + { + "left": "issue:47279", + "right": "issue:47419", + "accept": false, + "reason": "Separate daily triage artifacts from the same reporting stream." + }, + { + "left": "issue:47406", + "right": "issue:47419", + "accept": false, + "reason": "Same triage template, but distinct day-specific issues." + }, + { + "left": "issue:47279", + "right": "issue:47442", + "accept": false, + "reason": "No indication these are the same concrete bug or change." + }, + { + "left": "issue:47227", + "right": "issue:47442", + "accept": false, + "reason": "Later report in the same series, but not a duplicate of the earlier one." + }, + { + "left": "issue:47390", + "right": "issue:47419", + "accept": false, + "reason": "Recurring integration triage entries; different dates and likely different failures." + }, + { + "left": "issue:47390", + "right": "issue:47406", + "accept": false, + "reason": "High similarity comes from the common triage template, not a shared bug." + }, + { + "left": "issue:47390", + "right": "issue:47442", + "accept": false, + "reason": "Distinct daily reports; do not collapse them into one issue." + }, + { + "left": "issue:47406", + "right": "issue:47442", + "accept": false, + "reason": "Same recurring triage series, but separate report instances." + }, + { + "left": "issue:47377", + "right": "issue:47406", + "accept": false, + "reason": "Different dates in the same recurring triage stream." + }, + { + "left": "issue:47279", + "right": "issue:47406", + "accept": false, + "reason": "Template similarity only; no concrete duplicate evidence." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 25, + "component_count": 28, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46906-23", + "nodes": [ + "issue:47227", + "issue:47377", + "issue:47390" + ], + "soft_pairs": [ + "issue:47377|issue:47390", + "issue:47227|issue:47390" + ], + "prepared_review_unit_hash": "f9f22b33a22607112efa27ee138c67acb3c96440ed100fcb1347fc572d15cfe2", + "component_budget": { + "node_count": 23, + "item_count": 23, + "soft_pair_count": 74, + "serialized_chars": 39466, + "estimated_input_tokens": 9867, + "estimated_eval_tokens": 19990 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2318, + "estimated_input_tokens": 580, + "estimated_eval_tokens": 1416 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These look like recurring daily triage issues for serge integration failures, not the same underlying bug. They share a template-like title but have different dates and no linking evidence.", + "confidence": 0.76, + "canonical_issue_reason": "Issue #47377 is the best representative because it has the highest inbound reference count and matches the recurring triage pattern.", + "canonical_pr_reason": null, + "best_issue_reason": "#47377 is the strongest issue to keep as the cluster representative; the set appears to be multiple dated triage reports rather than duplicates of one defect.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:47377", + "right": "issue:47390", + "accept": false, + "reason": "Same triage template, but the dated titles indicate separate daily reports with no evidence of a single shared failure." + }, + { + "left": "issue:47227", + "right": "issue:47390", + "accept": false, + "reason": "Highly similar wording, but these are date-specific triage issues and not clearly the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 26, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46983-2", + "nodes": [ + "issue:46983", + "issue:46984" + ], + "soft_pairs": [ + "issue:46983|issue:46984" + ], + "prepared_review_unit_hash": "dd0d5445b1451e992dea6e9d12c26018b3c612d4cb70f30963dff3aabb9b1563", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1492, + "estimated_input_tokens": 373, + "estimated_eval_tokens": 1002 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1492, + "estimated_input_tokens": 373, + "estimated_eval_tokens": 1002 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two closed issues are highly similar requests for RWKV7 model support, likely describing the same feature request, but the evidence is only soft similarity and does not justify a definitive duplicate call. No PRs are present in the cluster.", + "confidence": 0.88, + "canonical_issue_reason": "#46984 is the better canonical issue because it has the same title as #46983 and slightly more discussion activity, making it the more established representative thread.", + "canonical_pr_reason": null, + "best_issue_reason": "#46984 is the strongest representative for the cluster since it matches the shared RWKV7 support request exactly and has a bit more activity.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46983", + "right": "issue:46984", + "accept": false, + "reason": "The issues are very similar and appear to request the same RWKV7 support, but the packet only provides soft-similarity evidence and deterministic_accept is false, so this should be treated as likely related rather than a definitive duplicate." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary is mostly grounded, but the verdict is a bit too strong. The packet only provides soft-similarity evidence between the two issues, and deterministic_accept is false, so calling them definitive duplicates overstates the evidence. It would be safer to say they are highly similar / likely the same request for RWKV7 support. The canonical issue choice is reasonable based on slightly higher discussion activity, and the note that there are no PRs is supported." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 27, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47007-2", + "nodes": [ + "issue:47007", + "issue:47015" + ], + "soft_pairs": [ + "issue:47007|issue:47015" + ], + "prepared_review_unit_hash": "1022a6665bf08a9a4f44330a44d99756406279f02c764c91bb4f6866dc70a13a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1626, + "estimated_input_tokens": 407, + "estimated_eval_tokens": 1070 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1626, + "estimated_input_tokens": 407, + "estimated_eval_tokens": 1070 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both issues are broad, essay-like posts about AI-driven vulnerability discovery, but they do not describe a concrete shared bug or code change. They are thematically related rather than clear duplicates, so the soft pair should be rejected.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 47015 is the more general, umbrella-style title and reads like the cleaner representative of the topic cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 47015 is the best representative because it is broader and more directly names the underlying subject matter without the extra 'lessons from using advanced LLMs' framing.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:47007", + "right": "issue:47015", + "accept": false, + "reason": "The titles overlap on AI-based vulnerability discovery, but there is no concrete bug, code path, or explicit shared change. These look like related high-level writeups, not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 28, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47138-2", + "nodes": [ + "issue:47138", + "issue:47139" + ], + "soft_pairs": [ + "issue:47138|issue:47139" + ], + "prepared_review_unit_hash": "16883a79be63f75afdb2e5e8500d5b1b44df2c5d04bc4c09e79d439424a74ec3", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1591, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1591, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two items are effectively duplicate issues with identical titles and near-identical bodies about the same /v1 base_url routing bug in transformers chat management endpoints.", + "confidence": 0.99, + "canonical_issue_reason": "Issue 47138 is the better canonical issue because it was created slightly earlier and has more inbound references, while describing the same problem as 47139.", + "canonical_pr_reason": null, + "best_issue_reason": "47138 is the strongest single issue representative of this bug due to earliest creation time and slightly higher discussion/reference activity.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:47138", + "right": "issue:47139", + "accept": true, + "reason": "Same title, same apparent body length/content, and the same underlying routing bug; these are duplicates of one issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet: both issues have identical titles, identical body length, and a soft-similarity score of 1.0, with only a small difference in creation time and inbound references. The duplicate verdict and canonical choice are conservative enough for the available evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 4, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-82", + "nodes": [ + "pull_request:43251", + "pull_request:43254", + "pull_request:44775", + "pull_request:45157", + "pull_request:46196", + "pull_request:46266", + "pull_request:46395", + "pull_request:46836", + "pull_request:46903", + "pull_request:46912", + "pull_request:46957", + "pull_request:46958", + "pull_request:46995", + "pull_request:47003", + "pull_request:47077", + "pull_request:47078", + "pull_request:47090", + "pull_request:47093", + "pull_request:47105", + "pull_request:47108", + "pull_request:47129", + "pull_request:47163", + "pull_request:47166", + "pull_request:47173", + "pull_request:47179", + "pull_request:47202", + "pull_request:47204", + "pull_request:47210", + "pull_request:47212", + "pull_request:47216", + "pull_request:47243", + "pull_request:47268", + "pull_request:47275", + "pull_request:47284", + "pull_request:47291", + "pull_request:47292", + "pull_request:47301", + "pull_request:47336", + "pull_request:47357", + "pull_request:47360", + "pull_request:47374", + "pull_request:47383", + "pull_request:47384", + "pull_request:47395", + "pull_request:47420", + "pull_request:47422", + "pull_request:47446", + "pull_request:47447" + ], + "soft_pairs": [ + "pull_request:43251|pull_request:43254", + "pull_request:46903|pull_request:47090", + "pull_request:47090|pull_request:47395", + "pull_request:46903|pull_request:46958", + "pull_request:46903|pull_request:47395", + "pull_request:47268|pull_request:47284", + "pull_request:47173|pull_request:47284", + "pull_request:47202|pull_request:47204", + "pull_request:47204|pull_request:47374", + "pull_request:47284|pull_request:47292", + "pull_request:47108|pull_request:47202", + "pull_request:47077|pull_request:47078", + "pull_request:47204|pull_request:47243", + "pull_request:47202|pull_request:47360", + "pull_request:47108|pull_request:47360", + "pull_request:47105|pull_request:47129", + "pull_request:47077|pull_request:47212", + "pull_request:46266|pull_request:46957", + "pull_request:47301|pull_request:47360", + "pull_request:47360|pull_request:47447", + "pull_request:47360|pull_request:47383", + "pull_request:47384|pull_request:47447", + "pull_request:44775|pull_request:47275", + "pull_request:47166|pull_request:47173", + "pull_request:47291|pull_request:47384", + "pull_request:47077|pull_request:47275", + "pull_request:47336|pull_request:47446", + "pull_request:47275|pull_request:47301", + "pull_request:47422|pull_request:47446", + "pull_request:47216|pull_request:47275", + "pull_request:46912|pull_request:47284", + "pull_request:47163|pull_request:47284", + "pull_request:46196|pull_request:47202", + "pull_request:46912|pull_request:47179", + "pull_request:47204|pull_request:47357", + "pull_request:46196|pull_request:47210", + "pull_request:46836|pull_request:47202", + "pull_request:47003|pull_request:47204", + "pull_request:47093|pull_request:47275", + "pull_request:45157|pull_request:47093", + "pull_request:46995|pull_request:47163", + "pull_request:47163|pull_request:47173", + "pull_request:47108|pull_request:47420", + "pull_request:45157|pull_request:46395", + "pull_request:46836|pull_request:47275" + ], + "prepared_review_unit_hash": "443f7bb7a59e01c9385e3c858c203d0d89d3c3072b4513b98c4c3bab5d827a39", + "component_budget": { + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 45, + "serialized_chars": 125413, + "estimated_input_tokens": 31354, + "estimated_eval_tokens": 62964 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous cluster with several unrelated PRs. Only a few edges look like true duplicates: the fixed_cross_entropy kwargs fixes, and the CohereASR/PPFormulaNet training-loss double-shift fixes. Most other pairs are same broad area or just docs/tests, not the same underlying change.", + "confidence": 0.47, + "canonical_issue_reason": null, + "canonical_pr_reason": "If forced to pick one PR representative, 47090 is the strongest canonical candidate among the genuine duplicates: it addresses the same concrete double-shift loss bug, covers the affected ASR-related models more broadly, and has the clearest issue linkage/tests.", + "best_issue_reason": null, + "best_pr_reason": "47090 is the best standalone PR in the cluster because it is a concrete code fix for a real bug, not just docs/tests, and it is the most complete version of the actual duplicate group here.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Same underlying cross-entropy kwargs bug in the same helper; both patch the same code path and could be merged into one PR." + }, + { + "left": "pull_request:46903", + "right": "pull_request:47090", + "accept": true, + "reason": "Same training-loss double-shift bug in the same model family; 47090 is the broader version of the same fix." + }, + { + "left": "pull_request:47090", + "right": "pull_request:47395", + "accept": false, + "reason": "Different models and different loss paths; both are double-shift fixes, but not the same code-path problem." + }, + { + "left": "pull_request:46903", + "right": "pull_request:46958", + "accept": true, + "reason": "Same CohereASR/PPFormulaNet double-shift loss bug; these are alternative implementations of the same fix." + }, + { + "left": "pull_request:46903", + "right": "pull_request:47395", + "accept": false, + "reason": "Different model family and unrelated loss logic; shared theme is too broad to treat as duplicate." + }, + { + "left": "pull_request:47268", + "right": "pull_request:47284", + "accept": false, + "reason": "Unrelated failing tests in different model families; no shared underlying bug." + }, + { + "left": "pull_request:47173", + "right": "pull_request:47284", + "accept": false, + "reason": "Both are test-only changes, but for different models and different failures." + }, + { + "left": "pull_request:47202", + "right": "pull_request:47204", + "accept": false, + "reason": "Both touch Docker/CI, but one drops a builder and the other adds exporter dependencies; not the same change." + }, + { + "left": "pull_request:47204", + "right": "pull_request:47374", + "accept": false, + "reason": "Executorch Docker test deps versus exporter docs; related ecosystem, but different artifacts and goals." + }, + { + "left": "pull_request:47284", + "right": "pull_request:47292", + "accept": false, + "reason": "Different model test fixes with no shared code path or bug." + }, + { + "left": "pull_request:47108", + "right": "pull_request:47202", + "accept": false, + "reason": "Doc-builder pin comment cleanup is unrelated to removing the CPU/light docker builder." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47078", + "accept": false, + "reason": "Both are Japanese doc edits, but one is a translation PR and the other is wording cleanup; not the same underlying change." + }, + { + "left": "pull_request:47204", + "right": "pull_request:47243", + "accept": false, + "reason": "Exporter Docker image dependencies versus exporter code fixes; related area, but not one mergeable bug fix." + }, + { + "left": "pull_request:47202", + "right": "pull_request:47360", + "accept": false, + "reason": "Different CI workflow changes with no common underlying fix." + }, + { + "left": "pull_request:47108", + "right": "pull_request:47360", + "accept": false, + "reason": "Unrelated documentation comment tweak versus OTEL CI workflow change." + }, + { + "left": "pull_request:47105", + "right": "pull_request:47129", + "accept": false, + "reason": "VITS test re-enable versus RT-DETR converter fix; unrelated model areas." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47212", + "accept": false, + "reason": "Japanese translation edits versus kv-cache docs updates; same language only, not same change." + }, + { + "left": "pull_request:46266", + "right": "pull_request:46957", + "accept": false, + "reason": "Different subsystems: checkpoint conversion mapping/model support versus generation kernel optimization." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47360", + "accept": false, + "reason": "Assisted-decoding docs versus CI observability workflow; no shared bug or change." + }, + { + "left": "pull_request:47360", + "right": "pull_request:47447", + "accept": false, + "reason": "CI workflow changes versus speculative-decoding cache rollback; unrelated." + }, + { + "left": "pull_request:47360", + "right": "pull_request:47383", + "accept": false, + "reason": "CI workflow change versus Kimi integration tests; not the same artifact." + }, + { + "left": "pull_request:47384", + "right": "pull_request:47447", + "accept": false, + "reason": "Gemma 4 heterogeneous config change versus cache rollback feature; different code paths." + }, + { + "left": "pull_request:44775", + "right": "pull_request:47275", + "accept": false, + "reason": "Both are docs/model additions, but for different subjects and no shared underlying change." + }, + { + "left": "pull_request:47166", + "right": "pull_request:47173", + "accept": false, + "reason": "Different test suites and failures; no evidence of the same bug." + }, + { + "left": "pull_request:47291", + "right": "pull_request:47384", + "accept": false, + "reason": "Both touch generation/configuration, but one deprecates a config path and the other changes heterogeneous per-layer config." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47275", + "accept": false, + "reason": "Japanese docs translation versus adding a new model; unrelated." + }, + { + "left": "pull_request:47336", + "right": "pull_request:47446", + "accept": false, + "reason": "Offloading save bugfix versus typo/comment cleanup; not duplicate changes." + }, + { + "left": "pull_request:47275", + "right": "pull_request:47301", + "accept": false, + "reason": "New model support versus assisted-decoding docs; unrelated." + }, + { + "left": "pull_request:47422", + "right": "pull_request:47446", + "accept": false, + "reason": "CI/docker pinning fix versus comment typo cleanup; different scope and code path." + }, + { + "left": "pull_request:47216", + "right": "pull_request:47275", + "accept": false, + "reason": "Quantization integration work versus a new vision model; unrelated." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47284", + "accept": false, + "reason": "RoPE test improvements versus MIMO v2 flash test fixes; unrelated." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47284", + "accept": false, + "reason": "Different XPU-related test changes for different components; not the same bug." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47202", + "accept": false, + "reason": "SDPA dispatch model support versus docker builder removal; unrelated." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47179", + "accept": false, + "reason": "RoPE test changes versus Granite SWA model support; different goals and code paths." + }, + { + "left": "pull_request:47204", + "right": "pull_request:47357", + "accept": false, + "reason": "Exporter test dependencies versus FSDP CI/save changes; not the same underlying issue." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47210", + "accept": false, + "reason": "Both are attention-related, but one adds SDPA dispatch and the other routes packed rows through varlen attention; different implementation problems." + }, + { + "left": "pull_request:46836", + "right": "pull_request:47202", + "accept": false, + "reason": "Unlimited OCR model support versus docker builder removal; unrelated." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47204", + "accept": false, + "reason": "OpenVINO exporter work versus ExecuTorch/torch-light dependency changes; different exporter targets." + }, + { + "left": "pull_request:47093", + "right": "pull_request:47275", + "accept": false, + "reason": "OpenPangu model addition versus EfficientViT-SAM model addition; same broad model-support theme, but distinct architectures and PRs." + }, + { + "left": "pull_request:45157", + "right": "pull_request:47093", + "accept": false, + "reason": "Prism GGUF support versus new model addition; unrelated." + }, + { + "left": "pull_request:46995", + "right": "pull_request:47163", + "accept": false, + "reason": "Availability check for flash linear attention versus paged-attention tests; different layers of the stack." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47173", + "accept": false, + "reason": "Both are XPU-related tests, but they target different failures and code paths." + }, + { + "left": "pull_request:47108", + "right": "pull_request:47420", + "accept": false, + "reason": "Doc-builder comment formatting versus tokenization special-token return cleanup; unrelated." + }, + { + "left": "pull_request:45157", + "right": "pull_request:46395", + "accept": false, + "reason": "Prism GGUF dequantization support versus Param2MoE model support; different features." + }, + { + "left": "pull_request:46836", + "right": "pull_request:47275", + "accept": false, + "reason": "Unlimited OCR support versus EfficientViT-SAM support; unrelated model additions." + }, + { + "left": "pull_request:46957", + "right": "pull_request:47210", + "accept": false, + "reason": "Finegrained FP8 decode optimization versus packed SDPA varlen attention; different generation optimizations." + }, + { + "left": "pull_request:47275", + "right": "pull_request:47301", + "accept": false, + "reason": "Model addition versus MTP documentation; no shared underlying change." + }, + { + "left": "pull_request:47422", + "right": "pull_request:47446", + "accept": false, + "reason": "Docker/CI pinning and RNNT fix versus typo-only comment cleanup; not duplicates." + }, + { + "left": "pull_request:47384", + "right": "pull_request:47447", + "accept": false, + "reason": "Heterogeneous Gemma config change versus speculative decoding cache rollback; different code paths and features." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 32, + "review_unit_index": 2, + "review_unit_count": 4, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-82", + "nodes": [ + "pull_request:44517", + "pull_request:44775", + "pull_request:45157", + "pull_request:46196", + "pull_request:46266", + "pull_request:46267", + "pull_request:46509", + "pull_request:46604", + "pull_request:46658", + "pull_request:46836", + "pull_request:46912", + "pull_request:46957", + "pull_request:46995", + "pull_request:47003", + "pull_request:47077", + "pull_request:47078", + "pull_request:47082", + "pull_request:47093", + "pull_request:47108", + "pull_request:47113", + "pull_request:47143", + "pull_request:47163", + "pull_request:47166", + "pull_request:47173", + "pull_request:47175", + "pull_request:47179", + "pull_request:47204", + "pull_request:47210", + "pull_request:47216", + "pull_request:47243", + "pull_request:47275", + "pull_request:47292", + "pull_request:47294", + "pull_request:47295", + "pull_request:47301", + "pull_request:47304", + "pull_request:47306", + "pull_request:47320", + "pull_request:47322", + "pull_request:47336", + "pull_request:47357", + "pull_request:47367", + "pull_request:47383", + "pull_request:47384", + "pull_request:47420", + "pull_request:47422", + "pull_request:47446", + "pull_request:47447" + ], + "soft_pairs": [ + "pull_request:47204|pull_request:47295", + "pull_request:46912|pull_request:47082", + "pull_request:47163|pull_request:47384", + "pull_request:44775|pull_request:47301", + "pull_request:47301|pull_request:47447", + "pull_request:46196|pull_request:46995", + "pull_request:47301|pull_request:47384", + "pull_request:47216|pull_request:47384", + "pull_request:47243|pull_request:47304", + "pull_request:47292|pull_request:47446", + "pull_request:47322|pull_request:47446", + "pull_request:47077|pull_request:47301", + "pull_request:47243|pull_request:47447", + "pull_request:46604|pull_request:47357", + "pull_request:44775|pull_request:47077", + "pull_request:47166|pull_request:47292", + "pull_request:47294|pull_request:47320", + "pull_request:47166|pull_request:47446", + "pull_request:47367|pull_request:47420", + "pull_request:46267|pull_request:47383", + "pull_request:47113|pull_request:47422", + "pull_request:46912|pull_request:47173", + "pull_request:47163|pull_request:47357", + "pull_request:46509|pull_request:46995", + "pull_request:47093|pull_request:47143", + "pull_request:46995|pull_request:47143", + "pull_request:47357|pull_request:47383", + "pull_request:46266|pull_request:47301", + "pull_request:46196|pull_request:47179", + "pull_request:46266|pull_request:47383", + "pull_request:44775|pull_request:47304", + "pull_request:47093|pull_request:47163", + "pull_request:47179|pull_request:47210", + "pull_request:47082|pull_request:47210", + "pull_request:47294|pull_request:47447", + "pull_request:46509|pull_request:47357", + "pull_request:46604|pull_request:47003", + "pull_request:47082|pull_request:47108", + "pull_request:47108|pull_request:47301", + "pull_request:46266|pull_request:47216", + "pull_request:47078|pull_request:47275", + "pull_request:47082|pull_request:47275", + "pull_request:46266|pull_request:47336", + "pull_request:46995|pull_request:47301", + "pull_request:47175|pull_request:47179", + "pull_request:44775|pull_request:46836", + "pull_request:47163|pull_request:47179", + "pull_request:46604|pull_request:46658", + "pull_request:47082|pull_request:47383", + "pull_request:44517|pull_request:45157", + "pull_request:47210|pull_request:47304", + "pull_request:47275|pull_request:47336", + "pull_request:46957|pull_request:47216", + "pull_request:47216|pull_request:47336", + "pull_request:47306|pull_request:47420", + "pull_request:47204|pull_request:47422" + ], + "prepared_review_unit_hash": "eb67cf1c58525597716028d5cb5a640f694d072aeb96275ab8674cd47f68f9ea", + "component_budget": { + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 56, + "serialized_chars": 133672, + "estimated_input_tokens": 33418, + "estimated_eval_tokens": 67092 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: it mixes model additions, documentation updates, CI/test tweaks, quantization/export changes, and a few generation/runtime fixes. The soft links are superficial topical similarity rather than the same underlying bug or change, so none should be merged as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: the items are not a duplicate set and cover unrelated changes across docs, models, tests, CI, quantization, and generation.", + "best_issue_reason": null, + "best_pr_reason": "No global best PR; the cluster is too heterogeneous to safely pick one representative as the canonical duplicate target.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47204", + "right": "pull_request:47295", + "accept": false, + "reason": "Docker/export-test dependency changes vs a DiffusionGemma mask fix; unrelated code paths." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47082", + "accept": false, + "reason": "RoPE test data updates vs a new RoMa model implementation; not the same change." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47384", + "accept": false, + "reason": "Paged-attention XPU tests vs Gemma 4 heterogeneous config changes; different subsystems." + }, + { + "left": "pull_request:44775", + "right": "pull_request:47301", + "accept": false, + "reason": "Parallelism docs vs assisted-decoding MTP docs; distinct documentation topics." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47447", + "accept": false, + "reason": "MTP docs vs speculative-decoding cache rollback; different features and code paths." + }, + { + "left": "pull_request:46196", + "right": "pull_request:46995", + "accept": false, + "reason": "SDPA dispatch support vs flash-linear-attention availability; related attention theme but different backend support." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47384", + "accept": false, + "reason": "MTP docs vs Gemma 4 config refactor; unrelated changes." + }, + { + "left": "pull_request:47216", + "right": "pull_request:47384", + "accept": false, + "reason": "Compressed-tensors FP8 support vs Gemma 4 config changes; not the same bug or feature." + }, + { + "left": "pull_request:47243", + "right": "pull_request:47304", + "accept": false, + "reason": "ExecuTorch exporter fixes vs an Apple MPS docs cleanup; unrelated." + }, + { + "left": "pull_request:47292", + "right": "pull_request:47446", + "accept": false, + "reason": "XPU test expectation updates vs typo fixes in comments; different scopes entirely." + }, + { + "left": "pull_request:47322", + "right": "pull_request:47446", + "accept": false, + "reason": "Quark config API change vs typo-only cleanup; not duplicates." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47301", + "accept": false, + "reason": "Japanese DistilBERT docs vs MTP docs; unrelated documentation." + }, + { + "left": "pull_request:47243", + "right": "pull_request:47447", + "accept": false, + "reason": "ExecuTorch exporter fixes vs speculative-decoding cache rollback; different runtime areas." + }, + { + "left": "pull_request:46604", + "right": "pull_request:47357", + "accept": false, + "reason": "Mistral tekken tokenizer utilities vs FSDP CI/save changes; unrelated functionality." + }, + { + "left": "pull_request:44775", + "right": "pull_request:47077", + "accept": false, + "reason": "Parallelism docs vs Japanese DistilBERT docs; no shared underlying change." + }, + { + "left": "pull_request:47166", + "right": "pull_request:47292", + "accept": false, + "reason": "GPTQ test updates vs TIPSv2 DPT XPU test fixes; different model/test targets." + }, + { + "left": "pull_request:47294", + "right": "pull_request:47320", + "accept": false, + "reason": "Kernel-native causal_conv1d work vs response_schema deprecation; unrelated changes." + }, + { + "left": "pull_request:47166", + "right": "pull_request:47446", + "accept": false, + "reason": "GPTQ test changes vs comment typo cleanup; not the same issue." + }, + { + "left": "pull_request:47367", + "right": "pull_request:47420", + "accept": false, + "reason": "Pipeline device-placement docs vs tokenizer special-token builder cleanup; unrelated." + }, + { + "left": "pull_request:46267", + "right": "pull_request:47383", + "accept": false, + "reason": "Serve chat parsing update vs Kimi integration tests; different areas of the stack." + }, + { + "left": "pull_request:47113", + "right": "pull_request:47422", + "accept": false, + "reason": "Audio feature/proc requirement fixes vs Docker torch/torchaudio pinning; related domain but different problems." + }, + { + "left": "pull_request:47093", + "right": "pull_request:47143", + "accept": false, + "reason": "OpenPangu-V2 model support vs NeuCodec model support; separate model additions." + }, + { + "left": "pull_request:46995", + "right": "pull_request:47143", + "accept": false, + "reason": "Flash-linear-attention availability on MLU vs NeuCodec model support; unrelated." + }, + { + "left": "pull_request:47357", + "right": "pull_request:47383", + "accept": false, + "reason": "FSDP CI/save work vs Kimi integration tests; no shared code-path fix." + }, + { + "left": "pull_request:46266", + "right": "pull_request:47301", + "accept": false, + "reason": "Dino modernization work vs MTP docs; completely different topics." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47179", + "accept": false, + "reason": "SDPA dispatch support vs Granite-swa model support; unrelated." + }, + { + "left": "pull_request:46266", + "right": "pull_request:47383", + "accept": false, + "reason": "Dino modernization work vs Kimi integration tests; not duplicates." + }, + { + "left": "pull_request:44775", + "right": "pull_request:47304", + "accept": false, + "reason": "Parallelism docs vs MPS graph-cache docs; different documentation updates." + }, + { + "left": "pull_request:47093", + "right": "pull_request:47163", + "accept": false, + "reason": "OpenPangu-V2 model support vs XPU paged-attention tests; unrelated." + }, + { + "left": "pull_request:47179", + "right": "pull_request:47210", + "accept": false, + "reason": "Granite-swa model support vs packed SDPA varlen routing; different changes." + }, + { + "left": "pull_request:47082", + "right": "pull_request:47210", + "accept": false, + "reason": "RoMa model addition vs SDPA varlen packing; unrelated." + }, + { + "left": "pull_request:47294", + "right": "pull_request:47447", + "accept": false, + "reason": "causal_conv1d kernel integration vs speculative-decoding cache rollback; different functionality." + }, + { + "left": "pull_request:46509", + "right": "pull_request:47357", + "accept": false, + "reason": "Nemotron Omni model support vs FSDP CI/save changes; no shared bug or change." + }, + { + "left": "pull_request:46604", + "right": "pull_request:47003", + "accept": false, + "reason": "Mistral tekken utilities vs OpenVINO exporter; unrelated subsystems." + }, + { + "left": "pull_request:47082", + "right": "pull_request:47108", + "accept": false, + "reason": "RoMa model support vs doc-builder pin comment cleanup; not the same issue." + }, + { + "left": "pull_request:46266", + "right": "pull_request:47216", + "accept": false, + "reason": "Dino modernization work vs compressed-tensors FP8 support; unrelated model/quantization work." + }, + { + "left": "pull_request:47078", + "right": "pull_request:47275", + "accept": false, + "reason": "Japanese docs wording changes vs EfficientViT-SAM model support; different scopes." + }, + { + "left": "pull_request:47082", + "right": "pull_request:47336", + "accept": false, + "reason": "RoMa model support vs fully offloaded save bugfix; unrelated." + }, + { + "left": "pull_request:44517", + "right": "pull_request:45157", + "accept": false, + "reason": "Qwen3-TTS model support vs PrismML Bonsai support; separate model additions." + }, + { + "left": "pull_request:47210", + "right": "pull_request:47304", + "accept": false, + "reason": "Packed SDPA routing vs MPS graph-cache docs; unrelated." + }, + { + "left": "pull_request:47275", + "right": "pull_request:47336", + "accept": false, + "reason": "EfficientViT-SAM model support vs offloaded model saving bugfix; different code paths." + }, + { + "left": "pull_request:46957", + "right": "pull_request:47216", + "accept": false, + "reason": "Finegrained FP8 kernel/runtime changes vs compressed-tensors FP8 support; both quantization-themed but not the same fix." + }, + { + "left": "pull_request:47216", + "right": "pull_request:47336", + "accept": false, + "reason": "Compressed-tensors FP8 support vs offloaded save bugfix; unrelated." + }, + { + "left": "pull_request:47306", + "right": "pull_request:47420", + "accept": false, + "reason": "Neuron experimentation scripts vs tokenizer special-token return cleanup; unrelated." + }, + { + "left": "pull_request:47204", + "right": "pull_request:47422", + "accept": false, + "reason": "Exporter-test Docker image setup vs CI torch/torchaudio pinning; both CI-related but different problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 32, + "review_unit_index": 3, + "review_unit_count": 4, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-82", + "nodes": [ + "pull_request:41855", + "pull_request:42668", + "pull_request:43251", + "pull_request:43424", + "pull_request:43451", + "pull_request:43823", + "pull_request:43995", + "pull_request:44517", + "pull_request:44827", + "pull_request:45157", + "pull_request:45170", + "pull_request:45655", + "pull_request:46048", + "pull_request:46196", + "pull_request:46381", + "pull_request:46604", + "pull_request:46658", + "pull_request:46895", + "pull_request:46912", + "pull_request:46957", + "pull_request:47077", + "pull_request:47078", + "pull_request:47096", + "pull_request:47105", + "pull_request:47113", + "pull_request:47143", + "pull_request:47163", + "pull_request:47166", + "pull_request:47175", + "pull_request:47179", + "pull_request:47210", + "pull_request:47243", + "pull_request:47275", + "pull_request:47291", + "pull_request:47294", + "pull_request:47298", + "pull_request:47301", + "pull_request:47304", + "pull_request:47306", + "pull_request:47320", + "pull_request:47322", + "pull_request:47360", + "pull_request:47367", + "pull_request:47383", + "pull_request:47420", + "pull_request:47422", + "pull_request:47438", + "pull_request:47447" + ], + "soft_pairs": [ + "pull_request:47360|pull_request:47422", + "pull_request:46912|pull_request:47078", + "pull_request:47322|pull_request:47420", + "pull_request:47179|pull_request:47301", + "pull_request:47166|pull_request:47422", + "pull_request:47298|pull_request:47438", + "pull_request:47143|pull_request:47320", + "pull_request:47298|pull_request:47420", + "pull_request:46912|pull_request:47294", + "pull_request:47175|pull_request:47294", + "pull_request:47175|pull_request:47438", + "pull_request:46895|pull_request:47306", + "pull_request:46196|pull_request:47301", + "pull_request:47243|pull_request:47301", + "pull_request:47175|pull_request:47447", + "pull_request:46912|pull_request:47438", + "pull_request:47105|pull_request:47298", + "pull_request:47078|pull_request:47322", + "pull_request:45157|pull_request:46658", + "pull_request:44517|pull_request:47301", + "pull_request:47294|pull_request:47438", + "pull_request:47113|pull_request:47304", + "pull_request:46912|pull_request:47291", + "pull_request:47438|pull_request:47447", + "pull_request:43451|pull_request:44517", + "pull_request:44517|pull_request:47383", + "pull_request:44517|pull_request:47143", + "pull_request:47078|pull_request:47243", + "pull_request:47210|pull_request:47291", + "pull_request:46957|pull_request:47383", + "pull_request:46604|pull_request:47163", + "pull_request:46604|pull_request:47383", + "pull_request:46957|pull_request:47360", + "pull_request:47113|pull_request:47175", + "pull_request:46895|pull_request:47298", + "pull_request:46381|pull_request:47322", + "pull_request:46957|pull_request:47301", + "pull_request:47298|pull_request:47447", + "pull_request:47113|pull_request:47275", + "pull_request:46912|pull_request:47113", + "pull_request:47077|pull_request:47367", + "pull_request:47294|pull_request:47306", + "pull_request:46658|pull_request:46895", + "pull_request:47210|pull_request:47367", + "pull_request:47275|pull_request:47367", + "pull_request:43451|pull_request:46381", + "pull_request:46381|pull_request:47096", + "pull_request:41855|pull_request:43424", + "pull_request:41855|pull_request:43823", + "pull_request:42668|pull_request:43823", + "pull_request:43424|pull_request:47301", + "pull_request:43424|pull_request:47383", + "pull_request:43823|pull_request:45655", + "pull_request:44827|pull_request:47383", + "pull_request:44827|pull_request:45170", + "pull_request:43424|pull_request:44827", + "pull_request:43823|pull_request:43995", + "pull_request:43251|pull_request:43823", + "pull_request:44827|pull_request:46048", + "pull_request:43424|pull_request:43823", + "pull_request:43251|pull_request:44827", + "pull_request:44827|pull_request:45655" + ], + "prepared_review_unit_hash": "6ebc71068d5fd3cc36c24c76d39682166ec2953e0530b31f366173ad49b91d32", + "component_budget": { + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 62, + "serialized_chars": 131803, + "estimated_input_tokens": 32951, + "estimated_eval_tokens": 66158 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous cluster of mostly unrelated PRs: model additions, docs edits, test tweaks, exporter/runtime fixes, CI changes, and small bug fixes. The soft-similarity pairs are topical at best and do not look like duplicate PRs for the same underlying change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: the items span unrelated subsystems and change types, so there is no duplicate-equivalent PR to anchor the cluster.", + "best_issue_reason": null, + "best_pr_reason": "If a representative PR had to be chosen, 47298 is the broadest infrastructure-oriented fix, but it still does not match the rest of the cluster as a duplicate.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47360", + "right": "pull_request:47422", + "accept": false, + "reason": "Different changes: OTEL workflow routing vs torchaudio pinning and RNN-T/CI fixes." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47078", + "accept": false, + "reason": "RoPE test coverage changes are unrelated to Japanese documentation wording edits." + }, + { + "left": "pull_request:47322", + "right": "pull_request:47420", + "accept": false, + "reason": "Quark config-class rename is unrelated to a tokenizer special-token return cleanup." + }, + { + "left": "pull_request:47179", + "right": "pull_request:47301", + "accept": false, + "reason": "Granite model support and MTP docs are different features, not the same fix." + }, + { + "left": "pull_request:47166", + "right": "pull_request:47422", + "accept": false, + "reason": "Quantization test expectation updates are unrelated to torch/torchaudio pinning in CI." + }, + { + "left": "pull_request:47298", + "right": "pull_request:47438", + "accept": false, + "reason": "MI450 image/model failure fixes and less-materialized masks address different code paths." + }, + { + "left": "pull_request:47143", + "right": "pull_request:47320", + "accept": false, + "reason": "NeuCodec model support is unrelated to deprecating the old response schema." + }, + { + "left": "pull_request:47298", + "right": "pull_request:47420", + "accept": false, + "reason": "MI450 runtime fixes and tokenizer special-token cleanup are unrelated bugs." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47294", + "accept": false, + "reason": "RoPE test work and kernel-native causal_conv1d integration are unrelated changes." + }, + { + "left": "pull_request:47175", + "right": "pull_request:47294", + "accept": false, + "reason": "Olmo3 CI expectation updates do not overlap with kernel-native causal_conv1d support." + }, + { + "left": "pull_request:47175", + "right": "pull_request:47438", + "accept": false, + "reason": "Test-value updates for Olmo3 are unrelated to the DSA mask optimization." + }, + { + "left": "pull_request:46895", + "right": "pull_request:47306", + "accept": false, + "reason": "CohereASR double-shift fix and Neuron experiment scripts are different problems." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47301", + "accept": false, + "reason": "SDPA dispatch support for wav2vec2-family models is unrelated to MTP docs." + }, + { + "left": "pull_request:47243", + "right": "pull_request:47301", + "accept": false, + "reason": "Executorch exporter fixes and assisted-decoding docs are not the same code-path issue." + }, + { + "left": "pull_request:47175", + "right": "pull_request:47447", + "accept": false, + "reason": "Olmo3 test updates do not match speculative-decoding cache rollback changes." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47438", + "accept": false, + "reason": "RoPE test tuning and recurrent-attention mask reduction are unrelated." + }, + { + "left": "pull_request:47105", + "right": "pull_request:47298", + "accept": false, + "reason": "Re-enabling a VITS test is unrelated to MI450 model failure fixes." + }, + { + "left": "pull_request:47078", + "right": "pull_request:47322", + "accept": false, + "reason": "Japanese docs wording changes are unrelated to Quark config API changes." + }, + { + "left": "pull_request:45157", + "right": "pull_request:46658", + "accept": false, + "reason": "Prism GGUF support and Step-3.7 flash model work are different model/tooling efforts." + }, + { + "left": "pull_request:44517", + "right": "pull_request:47301", + "accept": false, + "reason": "Qwen3-TTS model support is unrelated to MTP documentation." + }, + { + "left": "pull_request:47294", + "right": "pull_request:47438", + "accept": false, + "reason": "Kernel-native causal_conv1d functions and DSA mask changes fix different mechanisms." + }, + { + "left": "pull_request:47113", + "right": "pull_request:47320", + "accept": false, + "reason": "Audio feature/processor requirement fixes are unrelated to response-schema deprecation." + }, + { + "left": "pull_request:44517", + "right": "pull_request:47383", + "accept": false, + "reason": "Qwen3-TTS model addition and Kimi integration tests are unrelated." + }, + { + "left": "pull_request:44517", + "right": "pull_request:47143", + "accept": false, + "reason": "Different model additions: Qwen3-TTS versus NeuCodec." + }, + { + "left": "pull_request:47078", + "right": "pull_request:47243", + "accept": false, + "reason": "Japanese docs wording edits are unrelated to Executorch exporter fixes." + }, + { + "left": "pull_request:47210", + "right": "pull_request:47291", + "accept": false, + "reason": "Packed SDPA routing and generation-config deprecation are unrelated features." + }, + { + "left": "pull_request:46957", + "right": "pull_request:47383", + "accept": false, + "reason": "Finegrained FP8 decode-path tuning is unrelated to Kimi integration tests." + }, + { + "left": "pull_request:46604", + "right": "pull_request:47163", + "accept": false, + "reason": "Mistral tekken tokenizer utilities and XPU paged-attention tests are unrelated." + }, + { + "left": "pull_request:46604", + "right": "pull_request:47383", + "accept": false, + "reason": "Mistral tekken utilities and Kimi integration tests have no shared code-path." + }, + { + "left": "pull_request:46957", + "right": "pull_request:47360", + "accept": false, + "reason": "Finegrained FP8 kernel changes and OTEL CI workflow changes are unrelated." + }, + { + "left": "pull_request:47113", + "right": "pull_request:47275", + "accept": false, + "reason": "Audio requirement fixes do not overlap with EfficientVit-SAM model support." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47113", + "accept": false, + "reason": "RoPE testing changes and audio backend requirement fixes are unrelated." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47367", + "accept": false, + "reason": "Japanese DistilBERT docs translation and pipeline device-placement clarification are unrelated docs updates." + }, + { + "left": "pull_request:47294", + "right": "pull_request:47306", + "accept": false, + "reason": "Kernel-native causal_conv1d functions do not match Neuron experiment scripts." + }, + { + "left": "pull_request:47210", + "right": "pull_request:47367", + "accept": false, + "reason": "Packed SDPA attention routing and pipeline device-placement docs are unrelated." + }, + { + "left": "pull_request:47275", + "right": "pull_request:47367", + "accept": false, + "reason": "EfficientVit-SAM model support is unrelated to pipeline device-placement clarification." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46381", + "accept": false, + "reason": "Molmo2 model addition and tokenizers version bump are unrelated." + }, + { + "left": "pull_request:46381", + "right": "pull_request:47096", + "accept": false, + "reason": "Tokenizer dependency/update work is unrelated to exportable Kimi support." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Mistral missing-methods work and Executorch dynamic-shape tests are different issues." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Mistral tokenizer method additions are unrelated to MobileLLM model support." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Processor-from-pretrained robustness and MobileLLM model addition are unrelated." + }, + { + "left": "pull_request:43424", + "right": "pull_request:47301", + "accept": false, + "reason": "Executorch dynamic-shape test coverage is unrelated to MTP docs." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "MobileLLM model support and Falcon output-interface refactor are different code paths." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Cross-entropy kwargs passthrough fix is unrelated to MobileLLM support." + }, + { + "left": "pull_request:44827", + "right": "pull_request:47383", + "accept": false, + "reason": "Mistral4 test fixes and Kimi integration tests are unrelated model families." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test fixes and layrnorm-to-layernorm renaming are unrelated." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "Executorch dynamic-shape export tests and Mistral4 test fixes are unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:46048", + "accept": false, + "reason": "MobileLLM model support and TDT loss kernel support are unrelated features." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Executorch export tests and MobileLLM model support do not address the same bug." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Cross-entropy kwargs handling and Mistral4 test updates are unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4 test fixes and config-resolution-order changes are unrelated issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 32, + "review_unit_index": 4, + "review_unit_count": 4, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-82", + "nodes": [ + "pull_request:40546", + "pull_request:43451", + "pull_request:43823", + "pull_request:43995", + "pull_request:44827", + "pull_request:45157", + "pull_request:45213", + "pull_request:46196", + "pull_request:46509", + "pull_request:46658", + "pull_request:46836", + "pull_request:46895", + "pull_request:46912", + "pull_request:46958", + "pull_request:47003", + "pull_request:47077", + "pull_request:47090", + "pull_request:47096", + "pull_request:47143", + "pull_request:47163", + "pull_request:47166", + "pull_request:47173", + "pull_request:47175", + "pull_request:47210", + "pull_request:47216", + "pull_request:47243", + "pull_request:47291", + "pull_request:47292", + "pull_request:47294", + "pull_request:47298", + "pull_request:47301", + "pull_request:47304", + "pull_request:47306", + "pull_request:47322", + "pull_request:47336", + "pull_request:47367", + "pull_request:47383", + "pull_request:47387" + ], + "soft_pairs": [ + "pull_request:43995|pull_request:45213", + "pull_request:44827|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:43823|pull_request:45213", + "pull_request:46895|pull_request:47090", + "pull_request:46958|pull_request:47090", + "pull_request:46895|pull_request:46958", + "pull_request:47173|pull_request:47292", + "pull_request:46912|pull_request:47175", + "pull_request:47096|pull_request:47383", + "pull_request:47301|pull_request:47304", + "pull_request:47077|pull_request:47304", + "pull_request:46658|pull_request:46836", + "pull_request:47210|pull_request:47301", + "pull_request:47163|pull_request:47166", + "pull_request:47003|pull_request:47243", + "pull_request:46509|pull_request:47143", + "pull_request:46509|pull_request:47163", + "pull_request:45157|pull_request:46836", + "pull_request:47163|pull_request:47292", + "pull_request:46836|pull_request:47387", + "pull_request:47291|pull_request:47322", + "pull_request:45157|pull_request:46509", + "pull_request:47301|pull_request:47336", + "pull_request:47216|pull_request:47383", + "pull_request:47336|pull_request:47383", + "pull_request:46509|pull_request:47387", + "pull_request:47304|pull_request:47367", + "pull_request:47301|pull_request:47383", + "pull_request:47216|pull_request:47301", + "pull_request:46912|pull_request:47298", + "pull_request:46509|pull_request:47301", + "pull_request:47175|pull_request:47298", + "pull_request:47163|pull_request:47210", + "pull_request:45157|pull_request:47387", + "pull_request:47294|pull_request:47298", + "pull_request:46658|pull_request:47387", + "pull_request:46196|pull_request:47143", + "pull_request:47143|pull_request:47163", + "pull_request:46658|pull_request:47143", + "pull_request:46658|pull_request:47322", + "pull_request:47143|pull_request:47387", + "pull_request:47143|pull_request:47322", + "pull_request:45157|pull_request:47143", + "pull_request:47301|pull_request:47367", + "pull_request:43451|pull_request:47291", + "pull_request:47166|pull_request:47383", + "pull_request:40546|pull_request:43451", + "pull_request:47306|pull_request:47367", + "pull_request:43451|pull_request:47143", + "pull_request:43451|pull_request:47163", + "pull_request:43451|pull_request:47383", + "pull_request:43451|pull_request:46509", + "pull_request:43451|pull_request:47096", + "pull_request:43451|pull_request:47003", + "pull_request:47003|pull_request:47096", + "pull_request:43451|pull_request:47322", + "pull_request:47096|pull_request:47322", + "pull_request:43451|pull_request:47306", + "pull_request:47096|pull_request:47306", + "pull_request:47003|pull_request:47306", + "pull_request:47096|pull_request:47387", + "pull_request:47003|pull_request:47387", + "pull_request:47096|pull_request:47143" + ], + "prepared_review_unit_hash": "cd92e37966c4c162686472aef59ca4a962a7ab140c18de0456f16e386debf2a5", + "component_budget": { + "node_count": 82, + "item_count": 82, + "soft_pair_count": 227, + "serialized_chars": 282833, + "estimated_input_tokens": 70709, + "estimated_eval_tokens": 141674 + }, + "budget": { + "node_count": 38, + "item_count": 38, + "soft_pair_count": 64, + "serialized_chars": 112164, + "estimated_input_tokens": 28041, + "estimated_eval_tokens": 56338 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a soft-similarity grab bag of unrelated PRs that share broad subsystems or test-style changes. The only true duplicate family is the CohereASR training-loss double-shift fix trio; everything else should be kept separate. I would use PR #47090 as the representative canonical PR for that duplicate family.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #47090 is the clearest final form of the only real duplicate set here: it targets the same CohereASR double-shift bug as #46895 and #46958, touches the same files, and states the intended label alignment most explicitly.", + "best_issue_reason": null, + "best_pr_reason": "PR #47090 is the best overall representative because it is the most explicit and polished version of the same concrete fix, while the other strong-similarity pairs in the cluster are unrelated false positives.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon output-collection refactor vs a model-creation skill PR; unrelated changes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 test/export fix vs model-creation skill; different model and codepath." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Different models and different fixes; only broad transformer-internals similarity." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "MobileLLM scaffold vs model-creation skill; not the same underlying change." + }, + { + "left": "pull_request:46895", + "right": "pull_request:47090", + "accept": true, + "reason": "Same CohereASR double-shift bug, same files, same issue target; 47090 is the later, clearer variant." + }, + { + "left": "pull_request:46958", + "right": "pull_request:47090", + "accept": true, + "reason": "Same CohereASR training-loss bug and codepath; 47090 is just the more explicit final fix." + }, + { + "left": "pull_request:46895", + "right": "pull_request:46958", + "accept": true, + "reason": "Both fix the same CohereASR loss double-shift bug in the same files." + }, + { + "left": "pull_request:47173", + "right": "pull_request:47292", + "accept": false, + "reason": "Both are XPU test expectation updates, but for different models/files and different failures." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47175", + "accept": false, + "reason": "RoPE-related test updates vs Olmo3 CI expectation tweaks; not the same bug." + }, + { + "left": "pull_request:47096", + "right": "pull_request:47383", + "accept": false, + "reason": "Kimi exporter/model work vs Kimi integration tests; related project area but different concrete changes." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47304", + "accept": false, + "reason": "MTP docs vs MPS graph-cache docs; unrelated documentation changes." + }, + { + "left": "pull_request:47077", + "right": "pull_request:47304", + "accept": false, + "reason": "Japanese DistilBERT docs vs MPS graph-cache docs; unrelated." + }, + { + "left": "pull_request:46658", + "right": "pull_request:46836", + "accept": false, + "reason": "Step3.7 Flash model conversion work vs Unlimited OCR model support; different model additions." + }, + { + "left": "pull_request:47210", + "right": "pull_request:47301", + "accept": false, + "reason": "Packed-SDPA runtime change vs MTP docs; not the same change." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47166", + "accept": false, + "reason": "Paged-attention XPU tests vs GPTQ/MXFP4 test tweaks; different test fixes." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47243", + "accept": false, + "reason": "OpenVINO exporter addition vs Executorch exporter fixes; different exporter backends." + }, + { + "left": "pull_request:47291", + "right": "pull_request:47322", + "accept": false, + "reason": "Generation-config deprecation vs AMD Quark config rename; unrelated quantization/config changes." + }, + { + "left": "pull_request:45157", + "right": "pull_request:46836", + "accept": false, + "reason": "Prism GGUF dequant support vs Unlimited OCR model support; unrelated model/format work." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47292", + "accept": false, + "reason": "Paged-attention XPU tests vs tipsv2_dpt XPU tests; distinct model failures." + }, + { + "left": "pull_request:46836", + "right": "pull_request:47387", + "accept": false, + "reason": "Unlimited OCR model support vs one-to-many conversion infrastructure; not the same change." + }, + { + "left": "pull_request:45157", + "right": "pull_request:46509", + "accept": false, + "reason": "Prism GGUF support vs Nemotron Omni model support; unrelated." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47336", + "accept": false, + "reason": "MTP docs vs fully offloaded model saving bugfix; unrelated." + }, + { + "left": "pull_request:47216", + "right": "pull_request:47383", + "accept": false, + "reason": "Compressed-tensors fp8 work vs Kimi integration tests; different changes." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47210", + "accept": false, + "reason": "Paged-attention XPU tests vs packed-SDPA runtime change; unrelated." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47322", + "accept": false, + "reason": "OpenVINO exporter vs Quark config rename; unrelated." + }, + { + "left": "pull_request:47096", + "right": "pull_request:47387", + "accept": false, + "reason": "Kimi exporter/model work vs one-to-many conversion support; different codepaths." + }, + { + "left": "pull_request:47304", + "right": "pull_request:47367", + "accept": false, + "reason": "MPS graph-cache docs vs pipeline device-placement clarification; unrelated docs." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47383", + "accept": false, + "reason": "MTP docs vs Kimi integration tests; unrelated." + }, + { + "left": "pull_request:47216", + "right": "pull_request:47301", + "accept": false, + "reason": "Compressed-tensors fp8 work vs MTP docs; unrelated." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47298", + "accept": false, + "reason": "RoPE test tuning vs MI450 image-decoding/model failures; different subsystems." + }, + { + "left": "pull_request:47163", + "right": "pull_request:47210", + "accept": false, + "reason": "Paged-attention test updates vs packed-SDPA runtime path; not the same change." + }, + { + "left": "pull_request:45157", + "right": "pull_request:47387", + "accept": false, + "reason": "Prism GGUF support vs one-to-many conversion support; unrelated." + }, + { + "left": "pull_request:47294", + "right": "pull_request:47298", + "accept": false, + "reason": "Kernel-native causal_conv1d support vs MI450 image-decoder fixes; different codepaths." + }, + { + "left": "pull_request:46658", + "right": "pull_request:47387", + "accept": false, + "reason": "Step3.7 Flash model work vs one-to-many conversion infrastructure; unrelated." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47143", + "accept": false, + "reason": "SDPA dispatch support vs NeuCodec model support; different model areas." + }, + { + "left": "pull_request:47143", + "right": "pull_request:47163", + "accept": false, + "reason": "NeuCodec model support vs paged-attention XPU tests; unrelated." + }, + { + "left": "pull_request:46658", + "right": "pull_request:47143", + "accept": false, + "reason": "Step3.7 Flash model work vs NeuCodec model support; unrelated." + }, + { + "left": "pull_request:46658", + "right": "pull_request:47322", + "accept": false, + "reason": "Step3.7 model work vs Quark config changes; unrelated." + }, + { + "left": "pull_request:47143", + "right": "pull_request:47387", + "accept": false, + "reason": "NeuCodec model support vs one-to-many conversion support; unrelated." + }, + { + "left": "pull_request:47143", + "right": "pull_request:47322", + "accept": false, + "reason": "NeuCodec model support vs Quark config rename; unrelated." + }, + { + "left": "pull_request:45157", + "right": "pull_request:47143", + "accept": false, + "reason": "Prism GGUF support vs NeuCodec model support; unrelated." + }, + { + "left": "pull_request:47301", + "right": "pull_request:47367", + "accept": false, + "reason": "MTP docs vs pipeline device-placement docs; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47291", + "accept": false, + "reason": "Molmo2 model addition vs generation-config deprecation; unrelated." + }, + { + "left": "pull_request:47166", + "right": "pull_request:47383", + "accept": false, + "reason": "GPTQ/MXFP4 test tweaks vs Kimi integration tests; different failures." + }, + { + "left": "pull_request:40546", + "right": "pull_request:43451", + "accept": false, + "reason": "VibeVoice and Molmo2 are separate model-addition PRs." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47143", + "accept": false, + "reason": "Molmo2 vs NeuCodec; different models and artifacts." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47163", + "accept": false, + "reason": "Molmo2 model addition vs paged-attention XPU tests; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47383", + "accept": false, + "reason": "Molmo2 vs Kimi integration tests; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46509", + "accept": false, + "reason": "Molmo2 vs Nemotron Omni; separate model additions." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47096", + "accept": false, + "reason": "Molmo2 vs Kimi exporter/model work; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47003", + "accept": false, + "reason": "Molmo2 vs OpenVINO exporter work; unrelated." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47096", + "accept": false, + "reason": "OpenVINO exporter addition vs Kimi exporter/model work; different exporter paths." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47306", + "accept": false, + "reason": "OpenVINO exporter vs Neuron experiments; unrelated." + }, + { + "left": "pull_request:47096", + "right": "pull_request:47322", + "accept": false, + "reason": "Kimi exporter/model work vs Quark config rename; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47306", + "accept": false, + "reason": "Molmo2 model addition vs Neuron experiments; unrelated." + }, + { + "left": "pull_request:47096", + "right": "pull_request:47387", + "accept": false, + "reason": "Kimi exporter/model work vs one-to-many conversion support; unrelated." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47387", + "accept": false, + "reason": "OpenVINO exporter vs one-to-many conversion support; unrelated." + }, + { + "left": "pull_request:47096", + "right": "pull_request:47143", + "accept": false, + "reason": "Kimi exporter/model work vs NeuCodec model support; unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 2, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40755-2", + "nodes": [ + "pull_request:40755", + "pull_request:41224" + ], + "soft_pairs": [ + "pull_request:40755|pull_request:41224" + ], + "prepared_review_unit_hash": "051aca41be6b9d5e295f3b976691a72e2b5372a742e36c36eea5db459af73c95", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two unrelated feature PRs: one adds TimesFM covariate forecasting, the other adds DINOv3 image classification support. They do not appear to fix the same code-path problem or represent the same change, so they should not be clustered as duplicates.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #41224 is the more self-contained change: it adds a specific DINOv3 image-classification capability with matching model, auto-model, docs, and tests. It is a cleaner representative of a single coherent feature.", + "best_issue_reason": null, + "best_pr_reason": "PR #41224 is the strongest standalone PR in the set because it has a narrower, clearly defined scope and a concrete end-user feature implementation. PR #40755 is also valid, but it is a different feature area and not a better cluster representative.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different model families and different features: TimesFM covariate forecasting vs DINOv3 image classification support. No shared bug or mergeable code-path overlap." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 3, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41356-2", + "nodes": [ + "pull_request:41356", + "pull_request:44339" + ], + "soft_pairs": [ + "pull_request:41356|pull_request:44339" + ], + "prepared_review_unit_hash": "aead1308c167b0bae6ff11bf9358340baa5e76f816618383418b505a10a9cd01", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two DEIMv2 pull requests target the same model family, but they are not the same change: PR 41356 is an early/basic model + image processor add, while PR 44339 is a much broader follow-up that adds loss code, auto-mappings, conversion tooling, modular files, and more integration work.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44339 is the more complete and representative DEIMv2 implementation; PR 41356 looks like an earlier partial version of the same model bring-up.", + "best_issue_reason": null, + "best_pr_reason": "PR 44339 is the strongest single artifact for the cluster because it subsumes the basic model work and includes the broader integration needed for a full Transformers addition.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41356", + "right": "pull_request:44339", + "accept": false, + "reason": "Same model area, but not the same concrete change: 44339 expands the implementation substantially and adds new code paths beyond the basic PR 41356. They look like related iterations, not mergeable duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 4, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42975-2", + "nodes": [ + "pull_request:42975", + "pull_request:42976" + ], + "soft_pairs": [ + "pull_request:42975|pull_request:42976" + ], + "prepared_review_unit_hash": "1a480bee2bdeb61df7a9ac95e70cf165714f147be5a33273d2cc2a566aa7d204", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests are broad GitHub Actions dependency refreshes, but they target different concrete action sets: PR 42975 is specifically about Node 24 compatibility for core workflow actions, while PR 42976 upgrades docker/build-push-action and a few other actions to latest versions. They overlap in affected workflows, but not in the same underlying change, so this is not a duplicate pair.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the better canonical PR because it has the clearer, narrower purpose and directly targets the Node 24 compatibility work across workflow actions.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the strongest representative of the cluster: the title matches the main maintenance goal, and the diff shows a coherent set of action-version bumps for compatibility rather than a mixed latest-version refresh.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Related workflow-maintenance changes, but they update different actions for different purposes. They do not appear to fix the same concrete code-path problem and would not naturally merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42978-4", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:45490", + "pull_request:46562" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:45490", + "pull_request:43448|pull_request:45490", + "pull_request:45490|pull_request:46562" + ], + "prepared_review_unit_hash": "9ed2f37d53bbabe4824b118702b97dda15d11aa072275159a623912728b148da", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 13047, + "estimated_input_tokens": 3262, + "estimated_eval_tokens": 6780 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 13047, + "estimated_input_tokens": 3262, + "estimated_eval_tokens": 6780 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are four separate model-addition pull requests (ViT NEPA, Molmo, CTSM, Step3p7). They overlap mainly in common integration files like docs and auto-mapping, but they implement different models and unrelated code paths.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45490 is the best representative of the cluster because it is a substantial standalone model addition and has the strongest shared integration footprint across the set, but it is still not the same change as the others.", + "best_issue_reason": null, + "best_pr_reason": "PR 45490 is the best single PR to keep as the cluster representative: it is fully self-contained, touches core integration files, and has the highest discussion activity among the candidates.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Different model additions (ViT NEPA vs Molmo). Shared docs/auto-registry files are expected for new model onboarding and do not indicate the same underlying change." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "These add different models and modify different model implementations; overlap is only in shared registry/docs plumbing." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo and CTSM are separate model integrations with distinct modeling, processing, and conversion code, so they are not mergeable as one PR." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46562", + "accept": false, + "reason": "CTSM and Step3p7 are unrelated model additions; shared auto-mapping and docs edits are boilerplate for new model support, not evidence of the same bug or feature." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 6, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43291-7", + "nodes": [ + "pull_request:43291", + "pull_request:43665", + "pull_request:44815", + "pull_request:46380", + "pull_request:46539", + "pull_request:46886", + "pull_request:46922" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43291|pull_request:46380", + "pull_request:43291|pull_request:46922", + "pull_request:43665|pull_request:46539", + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:46380", + "pull_request:43665|pull_request:46922", + "pull_request:44815|pull_request:46539", + "pull_request:44815|pull_request:46380", + "pull_request:44815|pull_request:46922", + "pull_request:46380|pull_request:46539", + "pull_request:46380|pull_request:46922", + "pull_request:46539|pull_request:46922", + "pull_request:46886|pull_request:46922" + ], + "prepared_review_unit_hash": "df7bce939e7bef926268497af0c6590c5265ed7bc24199864cbc80536409f5f1", + "component_budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 15, + "serialized_chars": 19604, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 + }, + "budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 15, + "serialized_chars": 19604, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly unrelated PRs spanning Whisper tests, CLIP/ViT tests, FP8 dequantization, generation watermarking, Phimoe CI, and YOLOS. The only true duplicate-like pair is the two YOLOS fixes, which both address the same device_map=\"auto\" device mismatch path in the same file.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "pull_request:46922 is the strongest representative because it includes the concrete YOLOS runtime fix plus the device-splitting safeguard, directly tied to the explicit issue target.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:46922 is the best overall PR: it has the clearest concrete bug fix, a direct issue target, and the broadest evidence of addressing the underlying YOLOS device mismatch.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Different models and different kinds of changes; both are test-oriented but do not fix the same underlying bug or change." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper tokenizer test cleanup is unrelated to the FP8 dequant/load-path fix." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46380", + "accept": false, + "reason": "Whisper test changes and generation/watermarking validation changes are unrelated." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46922", + "accept": false, + "reason": "Whisper tests and YOLOS device mismatch fixes are unrelated code paths." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46539", + "accept": false, + "reason": "Test edits in CLIP/ViT are unrelated to the Phimoe CI/integration test fix." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Different subsystems and different fixes; no shared underlying bug." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46380", + "accept": false, + "reason": "CLIP/ViT test changes are unrelated to generation config validation." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46922", + "accept": false, + "reason": "Different models and issues; no common bug or mergeable unified fix." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46539", + "accept": false, + "reason": "FP8 dequantization changes and Phimoe CI environment changes are unrelated." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46380", + "accept": false, + "reason": "FP8 quantization/dequant logic is unrelated to generation config watermarking validation." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46922", + "accept": false, + "reason": "No shared code path or underlying bug; FP8 fix and YOLOS device mismatch are unrelated." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46539", + "accept": false, + "reason": "Generation config validation and CI/device memory patching solve different problems." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46922", + "accept": false, + "reason": "Unrelated generation config/watermarking fix vs YOLOS runtime device mismatch." + }, + { + "left": "pull_request:46539", + "right": "pull_request:46922", + "accept": false, + "reason": "Phimoe CI test infrastructure changes are unrelated to YOLOS model device placement." + }, + { + "left": "pull_request:46886", + "right": "pull_request:46922", + "accept": true, + "reason": "Both PRs fix the same YOLOS device_map=\"auto\" device mismatch in the same model file, and the changes are plausibly mergeable into one PR." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the negative pair verdicts stay conservative. The only accepted soft-edge pair is the YOLOS pair, which is reasonably mergeable into one PR for the same bugfix given the shared file and issue target." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 7, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43488-3", + "nodes": [ + "pull_request:43488", + "pull_request:43532", + "pull_request:46747" + ], + "soft_pairs": [ + "pull_request:43488|pull_request:46747", + "pull_request:43488|pull_request:43532", + "pull_request:43532|pull_request:46747" + ], + "prepared_review_unit_hash": "d32c47f3cef7a6a16a99a100902f894dd5c04c28172130460a21e671dfbf0528", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 6347, + "estimated_input_tokens": 1587, + "estimated_eval_tokens": 3430 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 6347, + "estimated_input_tokens": 1587, + "estimated_eval_tokens": 3430 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster contains three unrelated draft/don't-merge PRs. They share some weak topical overlap around repo automation and tests, but they do not fix the same concrete bug or implement the same change.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: 43488 is a packaging/dependency/version-bump style change, 43532 is a workflow/tooling demo touching several CI files, and 46747 is a distinct security-gate workflow trigger change.", + "best_issue_reason": null, + "best_pr_reason": "PR 46747 is the most self-contained and narrowly scoped change, so it is the best representative of the cluster, even though it is not a duplicate of the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:46747", + "accept": false, + "reason": "Different changes: 43488 touches setup/dependency metadata and a test, while 46747 changes a CI workflow trigger/permissions path. No shared concrete bug or fix." + }, + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "43488 is about packaging/deps/versioning; 43532 edits workflow files plus testing/notification utilities. The shared test file is incidental and not the same underlying change." + }, + { + "left": "pull_request:43532", + "right": "pull_request:46747", + "accept": false, + "reason": "Both involve workflows, but they target different CI systems and problems: 43532 is a diff-display/comment-CI demo, while 46747 changes a security-gate workflow trigger. Not mergeable as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43612-2", + "nodes": [ + "pull_request:43612", + "pull_request:43613" + ], + "soft_pairs": [ + "pull_request:43612|pull_request:43613" + ], + "prepared_review_unit_hash": "edbe5673f6f45cf9fd77f733d67148e16fbf59186a99dba32cc077132ce7b050", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6082, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6082, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs add adjacent but distinct segmentation pipelines: one for promptable concept segmentation and one for promptable visual segmentation. They share scaffold/docs machinery, but target different tasks and model integrations, so they should not be deduplicated.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43613 is the broader, more substantial pipeline addition and covers more model/test integrations, so it is the better representative PR for this cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR #43613 is the stronger global representative because it spans a wider set of related model backends and pipeline/docs updates.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "They are related segmentation pipeline additions, but they implement different concrete tasks (concept vs visual segmentation) with different model/test targets, so they are not the same underlying change and would not naturally merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 9, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43651-2", + "nodes": [ + "pull_request:43651", + "pull_request:43663" + ], + "soft_pairs": [ + "pull_request:43651|pull_request:43663" + ], + "prepared_review_unit_hash": "40bcf4102ef83ec08d7f44e1a05ad1484dd65935e2cd13a707956413c98a980f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs touch `src/transformers/trainer.py` and add custom-trainer extensibility, but they address different code paths: one changes gradient-accumulation loss scaling, the other refactors signature-column filtering.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "If a representative PR is needed, 43651 is the more concrete standalone behavior change; 43663 is a separate hook/refactor for column filtering, not the same fix.", + "best_issue_reason": null, + "best_pr_reason": "43651 is the stronger standalone candidate in this cluster, but it should not be treated as a duplicate of 43663.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different underlying Trainer changes: 43651 alters loss scaling during gradient accumulation, while 43663 adds an overridable signature-column helper for dataset filtering. Same file/subsystem is not enough to merge them." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 10, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43656-2", + "nodes": [ + "pull_request:43656", + "pull_request:43842" + ], + "soft_pairs": [ + "pull_request:43656|pull_request:43842" + ], + "prepared_review_unit_hash": "5017c0ee5e6fac8fe6ea0d94726537cfd095ac8b431231feb223224a1dbc12a8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3662, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3662, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs touch the same CLI server file and involve TypeAdapter-related typing, but they fix different NameError paths: one is a broader annotation cleanup, the other is a specific pydantic-absent import/annotation issue. They should not be merged as duplicates.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "43842 is the more concrete and narrowly scoped fix: it directly targets the TypeAdapter NameError when pydantic is not installed by moving the import under TYPE_CHECKING and quoting the annotation.", + "best_issue_reason": null, + "best_pr_reason": "43842 is the best representative PR because it has a clear user-facing bug description, an explicit issue target, and a focused code change tied to one concrete failure mode.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": false, + "reason": "They overlap in file and theme, but fix different problems: 43656 is a broader typing/annotation change in serve.py, while 43842 specifically fixes TypeAdapter access when pydantic is missing. Not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 11, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43779-2", + "nodes": [ + "pull_request:43779", + "pull_request:43816" + ], + "soft_pairs": [ + "pull_request:43779|pull_request:43816" + ], + "prepared_review_unit_hash": "e7c674adcc031aa8883109a5dfc678b4407fb20b10764302092e384d031ba83a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests appear to implement the same SwanLab integration change: adding run id/resume support in `integration_utils.py` for issue #43698, with one variant also expanding the docs. They are strong duplicate PR candidates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43816 is the cleaner representative: it carries the same SwanLab id/resume support plus the accompanying documentation updates, making it the more complete version of the shared change.", + "best_issue_reason": null, + "best_pr_reason": "PR #43816 is the best global representative because it appears to include the full feature change and docs for SwanLab run id/resume support, while #43779 is the narrower implementation-focused variant.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43779", + "right": "pull_request:43816", + "accept": true, + "reason": "Same underlying SwanLab callback enhancement for id/resume support in the same file and same issue target; the doc differences do not change the core code-path change, so these are duplicate PRs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs target issue #43698, touch the same file, and the previews indicate the same SwanLab id/resume support with #43816 adding docs. The duplicate/mergeable verdict is conservative enough given the shared target and overlapping code change." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 12, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43996", + "pull_request:43997", + "pull_request:44007", + "pull_request:44010", + "pull_request:44013", + "pull_request:44017", + "pull_request:44018", + "pull_request:44019", + "pull_request:44024", + "pull_request:44044", + "pull_request:44056", + "pull_request:44059", + "pull_request:44066", + "pull_request:44068", + "pull_request:44071", + "pull_request:44072", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44085", + "pull_request:44086", + "pull_request:44098", + "pull_request:44101", + "pull_request:44116", + "pull_request:44129", + "pull_request:44154", + "pull_request:44161", + "pull_request:44722" + ], + "soft_pairs": [ + "pull_request:43996|pull_request:44085", + "pull_request:43996|pull_request:44044", + "pull_request:44066|pull_request:44085", + "pull_request:44007|pull_request:44072", + "pull_request:44072|pull_request:44722", + "pull_request:44066|pull_request:44072", + "pull_request:44013|pull_request:44044", + "pull_request:44066|pull_request:44086", + "pull_request:44066|pull_request:44071", + "pull_request:44018|pull_request:44068", + "pull_request:44018|pull_request:44066", + "pull_request:44066|pull_request:44068", + "pull_request:44019|pull_request:44722", + "pull_request:44019|pull_request:44085", + "pull_request:44086|pull_request:44722", + "pull_request:44018|pull_request:44019", + "pull_request:44019|pull_request:44071", + "pull_request:44068|pull_request:44116", + "pull_request:44018|pull_request:44116", + "pull_request:44068|pull_request:44722", + "pull_request:44071|pull_request:44722", + "pull_request:44018|pull_request:44085", + "pull_request:44018|pull_request:44086", + "pull_request:44071|pull_request:44085", + "pull_request:44068|pull_request:44086", + "pull_request:44085|pull_request:44086", + "pull_request:44059|pull_request:44086", + "pull_request:44098|pull_request:44101", + "pull_request:44059|pull_request:44722", + "pull_request:44018|pull_request:44071", + "pull_request:44017|pull_request:44073", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44722", + "pull_request:44068|pull_request:44071", + "pull_request:43997|pull_request:44073", + "pull_request:44017|pull_request:44071", + "pull_request:44071|pull_request:44086", + "pull_request:44073|pull_request:44101", + "pull_request:44073|pull_request:44161", + "pull_request:44056|pull_request:44101", + "pull_request:44059|pull_request:44071", + "pull_request:44018|pull_request:44059", + "pull_request:44059|pull_request:44066", + "pull_request:43997|pull_request:44017", + "pull_request:44101|pull_request:44161", + "pull_request:43997|pull_request:44056", + "pull_request:44071|pull_request:44076", + "pull_request:44074|pull_request:44101", + "pull_request:44056|pull_request:44161", + "pull_request:43997|pull_request:44018", + "pull_request:43997|pull_request:44161", + "pull_request:44074|pull_request:44161", + "pull_request:44018|pull_request:44161", + "pull_request:44010|pull_request:44071", + "pull_request:44018|pull_request:44129", + "pull_request:44010|pull_request:44073", + "pull_request:43997|pull_request:44074", + "pull_request:44010|pull_request:44018", + "pull_request:44071|pull_request:44129", + "pull_request:43997|pull_request:44129", + "pull_request:43996|pull_request:44161", + "pull_request:43996|pull_request:44017", + "pull_request:44010|pull_request:44044", + "pull_request:44044|pull_request:44074", + "pull_request:44066|pull_request:44722", + "pull_request:44085|pull_request:44722", + "pull_request:44085|pull_request:44129", + "pull_request:43996|pull_request:44007", + "pull_request:43996|pull_request:44129", + "pull_request:44044|pull_request:44129", + "pull_request:44044|pull_request:44154", + "pull_request:44024|pull_request:44076" + ], + "prepared_review_unit_hash": "00cb590ca571dfc345b7a06f808293d4a590f180d0b5b217eb3b00c70b7f9ca1", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84512, + "estimated_input_tokens": 21128, + "estimated_eval_tokens": 42512 + }, + "budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 72, + "serialized_chars": 76943, + "estimated_input_tokens": 19236, + "estimated_eval_tokens": 38728 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a family of model-specific PRs migrating output handling to the new capture_outputs/can_return_tuple pattern. The tracking issue is shared, but most PRs touch different model code paths, so only a few are true near-duplicates.", + "confidence": 0.79, + "canonical_issue_reason": "Issue #43979 is the umbrella tracker referenced by many PRs here and best matches the shared output-tracing migration theme.", + "canonical_pr_reason": "PR #44066 is the best representative: it has the explicit issue link, a coherent GPT-J/CodeGen refactor, and sits at the center of the GPT-J follow-up/overlap group.", + "best_issue_reason": "#43979 is the only clear cluster-level issue and captures the common migration goal across the PRs.", + "best_pr_reason": "#44066 is the cleanest and most representative PR for this cluster; later GPT-J variants look like overlapping follow-ups rather than a better canonical anchor.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different model families and code paths (CVT/FNet vs GPT-J); same issue theme is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Different models and unrelated forward/output implementations; not the same concrete change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Both target the same GPT-J output-tracing refactor; 44066 includes the same GPT-J path and could plausibly absorb 44085." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "Both are standardized-output refactors, but they touch different backbone models and distinct code paths." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "EfficientNet vs GPT-J/CodeGen are different model families and not a single mergeable change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "GPT-J/CodeGen versus EfficientNet: shared refactor theme only, not the same underlying code-path bug." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 and DeBERTa-v2 are unrelated implementations; no shared concrete change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J/CodeGen versus MGP-STR are different code paths and model families." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-J/CodeGen and MPT are separate model implementations; not a duplicate." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Same GPT-Neo refactor in the same file family; 44068 looks like a later iteration of the 44018 change." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo versus GPT-J/CodeGen are different model paths, so they should stay separate." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen and GPT-Neo are different code paths despite the same migration theme." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet output tracing and GPT-J/CodeGen output tracing are different model implementations." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet and GPT-J are unrelated code paths." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are different model families and cannot be merged as one PR." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are distinct models with separate output logic." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are different code paths; same output-tracing idea is too broad." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are unrelated implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert do not share the same concrete code-path problem." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo versus GPT-J/CodeGen are different model paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "MPT and GPT-J/CodeGen are unrelated implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo and GPT-J are different model families." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are separate code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44085", + "accept": false, + "reason": "MPT and GPT-J are different models; no shared concrete bug/change." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-2 and MGP-STR are different code paths." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert/XLM are different model families; same refactor theme is not enough." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-2 and GPT-J/CodeGen are separate model implementations." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "Segformer and VisualBert are unrelated code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "Segformer and LongT5 are different models; not the same concrete change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are distinct implementations." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are different code paths." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are different models." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet and LongT5 are separate code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet and GPT-Neo are unrelated model families." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are different code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 are unrelated implementations." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "CVT/FNet and LongT5 are different model paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "CVT/FNet and Segformer are separate models." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are different code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are unrelated implementations." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44161", + "accept": false, + "reason": "SqueezeBert and LongT5 do not share the same concrete change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different backbone families." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are unrelated model paths." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44071", + "accept": false, + "reason": "MPNet and MPT are different code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44059", + "accept": false, + "reason": "GPT-Neo and GPT-2 are separate model implementations." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBert and DeBERTa-v2 are distinct models." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTa-v2 and TextNet are unrelated code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Near-identical GPT-J/CodeGen output-tracing refactors on the same files; 44722 looks like a later overlapping iteration of 44066." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both touch GPT-J output tracing and appear to be overlapping versions of the same refactor, with 44722 adding the same GPT-J path plus CodeGen." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J and SpeechT5 are different model code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and ResNet/RT-DETR are different backbones; same issue target is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 are unrelated implementations." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTa-v2 and SpeechT5 are different code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa-v2 and VITS are separate models and not a single duplicate fix." + }, + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "FocalNet and ImageGPT are different model families; shared output-tracing pattern is too broad." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: the shared tracker issue #43979 is clearly central, and the canonical PR choice (#44066) is supported by its explicit issue link and its GPT-J/CodeGen overlap with later follow-ups. The soft-edge verdicts are mostly conservative, and the accepted pairs are the ones that are visibly same-family, same-file, or explicit overlap cases." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 12, + "component_count": 32, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43997", + "pull_request:44010", + "pull_request:44056", + "pull_request:44073", + "pull_request:44074", + "pull_request:44129" + ], + "soft_pairs": [ + "pull_request:44073|pull_request:44074", + "pull_request:44056|pull_request:44073", + "pull_request:44056|pull_request:44074", + "pull_request:43997|pull_request:44010", + "pull_request:44010|pull_request:44129" + ], + "prepared_review_unit_hash": "d286beac73f431011dbdc8c028d3609a9ffc5ad9cbd5a3ec35211edb75255083", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84512, + "estimated_input_tokens": 21128, + "estimated_eval_tokens": 42512 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 5, + "serialized_chars": 13069, + "estimated_input_tokens": 3268, + "estimated_eval_tokens": 6792 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a set of model-specific refactors to standardized output tracing/capture_outputs. They are stylistically similar, but they target different model implementations and code paths, so they should not be merged as duplicates.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single PR is a true duplicate of the others; if one must represent the cluster, 44129 is the most complete anchor because it has an explicit issue target and a concrete, self-contained SpeechT5 output-tracing refactor.", + "best_issue_reason": null, + "best_pr_reason": "44129 is the strongest representative to anchor discussion because it includes an explicit linked issue and a substantive output-capture change, while the other PRs are parallel model-specific refactors.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "Both are output-tracing refactors, but they apply to different models (VisualBert vs TextNet) and different files/code paths; same pattern, not the same change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "MPNet vs VisualBert are distinct model implementations; these are parallel decorator refactors rather than one underlying bug or mergeable single PR." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "Different models and different forward/output-capture behavior; the similarity is only the standardized refactor pattern." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are unrelated code paths; both migrate to output tracing, but they are not the same concrete fix." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "SqueezeBert and SpeechT5 have different model-specific output handling and touched code; they are thematically similar refactors, not duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43998-13", + "nodes": [ + "pull_request:43998", + "pull_request:43999", + "pull_request:44000", + "pull_request:44001", + "pull_request:44002", + "pull_request:44003", + "pull_request:44004", + "pull_request:44025", + "pull_request:44026", + "pull_request:44027", + "pull_request:44028", + "pull_request:44029", + "pull_request:44030" + ], + "soft_pairs": [ + "pull_request:43998|pull_request:44002", + "pull_request:43998|pull_request:44000", + "pull_request:43998|pull_request:44001", + "pull_request:43998|pull_request:44003", + "pull_request:43998|pull_request:44029", + "pull_request:43999|pull_request:44001", + "pull_request:43999|pull_request:44002", + "pull_request:43999|pull_request:44025", + "pull_request:43999|pull_request:44026", + "pull_request:43999|pull_request:44027", + "pull_request:44000|pull_request:44001", + "pull_request:44000|pull_request:44002", + "pull_request:44000|pull_request:44003", + "pull_request:44000|pull_request:44025", + "pull_request:44001|pull_request:44002", + "pull_request:44001|pull_request:44003", + "pull_request:44001|pull_request:44025", + "pull_request:44002|pull_request:44029", + "pull_request:44002|pull_request:44003", + "pull_request:44003|pull_request:44028", + "pull_request:44003|pull_request:44004", + "pull_request:43998|pull_request:44004", + "pull_request:44001|pull_request:44004", + "pull_request:44002|pull_request:44004", + "pull_request:44000|pull_request:44004", + "pull_request:44025|pull_request:44026", + "pull_request:44025|pull_request:44027", + "pull_request:44025|pull_request:44028", + "pull_request:44025|pull_request:44029", + "pull_request:44025|pull_request:44030", + "pull_request:44026|pull_request:44029", + "pull_request:44026|pull_request:44027", + "pull_request:44026|pull_request:44028", + "pull_request:44026|pull_request:44030", + "pull_request:44027|pull_request:44028", + "pull_request:44027|pull_request:44029", + "pull_request:44027|pull_request:44030", + "pull_request:44028|pull_request:44029", + "pull_request:44028|pull_request:44030", + "pull_request:44029|pull_request:44030" + ], + "prepared_review_unit_hash": "f5568d0a3af1f75c605ac987fb3eaedfb5665c7b6549a2643d5a2cfe9bc9479f", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40184, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40184, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a set of similarly named output-tracing refactors across unrelated model implementations, not duplicate fixes for one underlying bug. None of the soft edges look safe to merge as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: the items target different model codepaths, so they are only thematically similar.", + "best_issue_reason": null, + "best_pr_reason": "PR 44025 is the strongest representative because it updates two closely related depth-anything variants and has the broadest coherent scope in the set.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Same refactor theme, but different model codepaths (`timm_backbone` vs `upernet`)." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Both are output-tracing refactors, but they target unrelated models and cannot be merged as one fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Different architectures and codepaths; only the high-level refactor theme matches." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model families (`timm_backbone` vs `falcon_mamba`/`mamba`); not the same underlying change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Unrelated model implementations; same pattern, not the same bug or change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "Different model families; this is a shared refactor pattern, not a duplicate fix." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "Separate codepaths in different models; no single concrete change in common." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "Both touch output tracing, but on different model stacks and files." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "Different encoder-decoder vs vision model codepaths; not mergeable as one PR." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "Same theme, but distinct model implementations and fixes." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "These are unrelated model implementations; similarity is only in the refactor pattern." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model codepaths; not the same concrete bug/change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "Separate models with no shared fix beyond the generic output-tracing refactor." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Different model families and files; no single duplicate change." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "Only the refactor motif matches; the underlying implementations are unrelated." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model stacks and concrete codepaths." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "Different model families; not the same underlying fix." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "Separate models and implementations; no shared concrete bug." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model codepaths, despite similar output-tracing work." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "Different model families and implementation details; not duplicate PRs." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "These refactors touch different concrete model logic (`mamba` vs `codegen`)." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "Different models and change sets; same pattern only at a very high level." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "Unrelated codepaths; not the same underlying bug or change." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "Different model implementations; no safe duplicate merge." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Same refactor category, but different model-specific fixes." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Different model families (`depth_anything` vs `vision_encoder_decoder`); not mergeable as one PR." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "Different codepaths and model stacks; only the refactor theme is shared." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "Separate models with no shared concrete bug or implementation." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "Different models; too broad to treat as duplicate changes." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "Different architectures and files; not the same underlying change." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "Different model families and codepaths; similarity is only thematic." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "Both are encoder-decoder style refactors, but they target different model implementations." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "Unrelated models and concrete fixes." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "Different model stacks and no shared bug report or codepath." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "Different model implementations; not the same concrete PR." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "Separate codepaths in unrelated models." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "Different model families and fix scopes." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "Different concrete implementations; no shared duplicate change." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "Unrelated model codepaths, despite similar refactor wording." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "Different model families and changes; not a duplicate fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44662-2", + "nodes": [ + "pull_request:44662", + "pull_request:45401" + ], + "soft_pairs": [ + "pull_request:44662|pull_request:45401" + ], + "prepared_review_unit_hash": "b5285192e7f931d08fe358e603c4f6d6a7e0991dfd63db2e82c6cd4d156384d6", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6551, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6551, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two unrelated model-addition PRs that only share generic Transformers integration files; they do not appear to fix the same bug or implement the same change.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45401 is the more focused, explicitly scoped Voxtral TTS implementation with a linked tracking issue; PR 44662 is a separate PenguinVL model addition.", + "best_issue_reason": null, + "best_pr_reason": "45401 is the best representative of the cluster only in the sense that it is a self-contained model-support PR; 44662 is clearly a different model and change set.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Reject: PenguinVL implementation vs Voxtral TTS support are different models with different code paths and tests; overlapping registry/docs files are generic boilerplate, not evidence of the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 15, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44979-2", + "nodes": [ + "pull_request:44979", + "pull_request:45363" + ], + "soft_pairs": [ + "pull_request:44979|pull_request:45363" + ], + "prepared_review_unit_hash": "c1366eaa65a9610d3632ebd5f0bb09304a014af3ffedbc51c99a8231d0a34a26", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4407, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4407, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two related pull requests, but they are not duplicates: one introduces the core Module Fusion API, while the other adds a specific n-to-1 kernel fusion workflow built on top of that API.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #44979 is the broader foundational change: it adds the new `module_fusion.py` API and tests, which is the base abstraction other fusion features depend on.", + "best_issue_reason": null, + "best_pr_reason": "PR #44979 is the best cluster representative because it is the general reusable API change; #45363 is a narrower follow-on implementation using that API.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "They address different levels of the stack: #44979 adds the generic module-fusion API, while #45363 wires a specific `KernelConfig`-based n-to-1 kernel fusion path and example. Related, but not the same concrete change and not something that would plausibly be merged as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 16, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45086-2", + "nodes": [ + "pull_request:45086", + "pull_request:45317" + ], + "soft_pairs": [ + "pull_request:45086|pull_request:45317" + ], + "prepared_review_unit_hash": "88597779aa9af70ece8bb78010f1f716da9628895b06b7ec93d529048e0d3461", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs fix the same AttributeError in `_patch_mistral_regex` by switching from `tokenizer.backend_tokenizer.pre_tokenizer` to `tokenizer.pre_tokenizer`; 45317 is the more complete version because it adds test coverage.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45317 is the better canonical because it addresses the same bug as 45086 but also includes tests, making it the more complete change for the same code path.", + "best_issue_reason": null, + "best_pr_reason": "PR 45317 is the best PR because it contains the same fix plus validation, so it is the strongest representative of this duplicate cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying bug and same patch hunk in the same file; 45317 is essentially 45086 plus tests, so these are duplicate PRs for the same fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 17, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45115-2", + "nodes": [ + "pull_request:45115", + "pull_request:45569" + ], + "soft_pairs": [ + "pull_request:45115|pull_request:45569" + ], + "prepared_review_unit_hash": "51ef619e9ea5e5ecb8a90dd01c9c5f92244f8ddc442be054c262e89c03c723c9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6065, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6065, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related NemotronH PRs, but they target different changes: one is a refactor to make NemotronH inherit GraniteMoeHybrid and update conversion mapping, while the other is a broader model split into NemotronH Dense/Sparse with docs and auto-mapping updates. Overlap in shared files is not enough to treat them as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45569 is the better canonical cluster representative because it is broader and more user-visible, covering the model split, docs, and auto-mapping changes rather than just the narrower inheritance/conversion refactor.", + "best_issue_reason": null, + "best_pr_reason": "PR 45569 is the best single PR for this cluster because it encompasses the more complete NemotronH architecture update and superseding model-layout changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Reject: they touch the same NemotronH codebase, but 45115 focuses on GraniteMoeHybrid-compatible inheritance and checkpoint conversion renames, while 45569 adds NemotronH Dense/Sparse modules, docs, and auto-model wiring. Different concrete changes, not one mergeable fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 18, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45415-2", + "nodes": [ + "pull_request:45415", + "pull_request:45425" + ], + "soft_pairs": [ + "pull_request:45415|pull_request:45425" + ], + "prepared_review_unit_hash": "eef3625725490ba020a69a4131204754716c55c4490e74272144135a7347b79b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests are typing-related, but they address different changes: PR 45415 is a broad repo-wide type-checking pass across many `src/transformers/*.py` files, while PR 45425 introduces new typing protocols and targeted fixes around `modeling_utils`, PEFT, and flash-attention helpers. They overlap in theme and a few files, but not the same concrete change.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45415 is the broader umbrella change: it touches many files to make `src/transformers/*py` type-check cleanly, so it best represents the cluster\u2019s main typing effort.", + "best_issue_reason": null, + "best_pr_reason": "PR 45415 is the best single representative because it is the most comprehensive type-checking pass and captures the cluster\u2019s dominant scope better than the narrower follow-up typing PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Reject: both are typing cleanups, but 45415 is a broad type-checking sweep while 45425 adds specific typing protocols and targeted fixes; they are not the same underlying bug/change and would not naturally merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 19, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45548-2", + "nodes": [ + "pull_request:45548", + "pull_request:45662" + ], + "soft_pairs": [ + "pull_request:45548|pull_request:45662" + ], + "prepared_review_unit_hash": "7a5d9b49c516be318fec15776f73c20e9daedc36635711a4d47e21cac0d7fd76", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are about expert parallelism and share `modeling_utils.py`, but they fix different concrete failures: one is about preserving `distributed_config` / `has_ep` for DeepSpeed ZeRO-3 loading via accelerate, the other about preventing expert parameters from being overwritten by rank-0 broadcast in FSDP2. They are related but not duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: the two patches address different distributed-loading bugs, so neither subsumes the other.", + "best_issue_reason": null, + "best_pr_reason": "PR 45662 is the stronger representative for the cluster because it has the broader, more explicit distributed-state fix path and clearer failure mode, though it is not a duplicate of 45548.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Both involve EP/loading and `modeling_utils.py`, but 45548 fixes `distributed_config` serialization / `has_ep` for ZeRO-3 loading, while 45662 fixes expert parameters being overwritten during FSDP2 post-shard handling. Different root causes and different code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 20, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45639-2", + "nodes": [ + "pull_request:45639", + "pull_request:45645" + ], + "soft_pairs": [ + "pull_request:45639|pull_request:45645" + ], + "prepared_review_unit_hash": "972e221afea185d36a74a392ea27d770f12b5c91569b9b1caa40ebf154fc8658", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4672, + "estimated_input_tokens": 1168, + "estimated_eval_tokens": 2592 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4672, + "estimated_input_tokens": 1168, + "estimated_eval_tokens": 2592 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs target the same xdist-related captured_info/debug-log collision problem in `testing_utils`; PR #45645 appears to be the broader, more complete version because it adds the CI workflow change needed to consume multiple per-worker files.", + "confidence": 0.81, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45645 is the better canonical PR because it covers the core xdist-safe logging fix and also updates the workflow to aggregate the new per-worker artifacts, making it the more complete end-to-end change.", + "best_issue_reason": null, + "best_pr_reason": "PR #45645 is the strongest overall candidate: it addresses the bug in the producer and the downstream CI log collection path, so it is the most complete fix for the underlying xdist artifact collision issue.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Both PRs address the same captured_info/xdist collision problem and touch the same fix path in `testing_utils`; #45645 is essentially the broader superset that adds workflow handling, so they are close enough to treat as the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 21, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45737-2", + "nodes": [ + "pull_request:45737", + "pull_request:45738" + ], + "soft_pairs": [ + "pull_request:45737|pull_request:45738" + ], + "prepared_review_unit_hash": "cc2ade73a49ec6a8883a7a24c433988645588bcd049e8aad2bce56b1b24daa1c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same MusicgenMelody cache initialization bug in `forward()`: replacing `EncoderDecoderCache(DynamicCache, DynamicCache)` with a single `DynamicCache`. PR #45737 is the more complete duplicate because it bundles the fix with a regression test, while #45738 is the fix-only variant.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45737 is the best canonical PR because it contains the same code fix as #45738 plus a regression test, making it the most complete representation of the bug fix.", + "best_issue_reason": null, + "best_pr_reason": "PR #45737 is the strongest overall candidate: it fixes the concrete cache bug and adds a targeted regression test, whereas #45738 only changes the implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45737", + "right": "pull_request:45738", + "accept": true, + "reason": "Same underlying code-path bug in `musicgen_melody` cache setup; the diffs match on the core fix and the test-only addition in #45737 could plausibly live in the same PR as #45738." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet: both PRs make the same `musicgen_melody` cache initialization change, and #45737 only adds a regression test on top of that fix. The soft-edge accept is conservative enough because the diffs are mergeable into one PR for the same bugfix." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 22, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46126-2", + "nodes": [ + "pull_request:46126", + "pull_request:46269" + ], + "soft_pairs": [ + "pull_request:46126|pull_request:46269" + ], + "prepared_review_unit_hash": "3e41010c2264ff8682ca30d131e9807faae7e5fe30cbe15f43c4473c7ef46c0b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5525, + "estimated_input_tokens": 1382, + "estimated_eval_tokens": 3020 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5525, + "estimated_input_tokens": 1382, + "estimated_eval_tokens": 3020 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicates: one PR is a focused EP/trainer integration change, while the other is a broad distributed-branch base touching many unrelated distributed subsystems. Shared filenames alone are not enough to treat them as the same underlying change.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46126 is the more specific, self-contained change with a clear concrete purpose (EP + trainer integration) and is a better representative of the narrower topic in this cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR #46126 is the best PR candidate because it is the more focused implementation-level change; PR #46269 is too broad and platform-like to serve as a duplicate representative.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46126", + "right": "pull_request:46269", + "accept": false, + "reason": "Different scope and intent: #46126 adds EP routing/trainer integration around tensor-parallel/MoE code paths, while #46269 is a large distributed-branch base touching many files across FSDP, integrations, modeling, and CI. They do not look like the same concrete fix or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 23, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46138-10", + "nodes": [ + "pull_request:46138", + "pull_request:46432", + "pull_request:46501", + "pull_request:46502", + "pull_request:46503", + "pull_request:46504", + "pull_request:46554", + "pull_request:46558", + "pull_request:46563", + "pull_request:46716" + ], + "soft_pairs": [ + "pull_request:46138|pull_request:46554", + "pull_request:46138|pull_request:46558", + "pull_request:46138|pull_request:46563", + "pull_request:46138|pull_request:46716", + "pull_request:46138|pull_request:46504", + "pull_request:46432|pull_request:46503", + "pull_request:46432|pull_request:46501", + "pull_request:46432|pull_request:46502", + "pull_request:46432|pull_request:46554", + "pull_request:46432|pull_request:46558", + "pull_request:46501|pull_request:46503", + "pull_request:46501|pull_request:46502", + "pull_request:46501|pull_request:46504", + "pull_request:46501|pull_request:46554", + "pull_request:46502|pull_request:46503", + "pull_request:46502|pull_request:46504", + "pull_request:46502|pull_request:46554", + "pull_request:46503|pull_request:46504", + "pull_request:46503|pull_request:46554", + "pull_request:46504|pull_request:46554", + "pull_request:46504|pull_request:46558", + "pull_request:46554|pull_request:46558", + "pull_request:46554|pull_request:46563", + "pull_request:46554|pull_request:46716", + "pull_request:46558|pull_request:46563", + "pull_request:46558|pull_request:46716", + "pull_request:46563|pull_request:46716", + "pull_request:46504|pull_request:46563", + "pull_request:46432|pull_request:46563", + "pull_request:46504|pull_request:46716", + "pull_request:46432|pull_request:46716" + ], + "prepared_review_unit_hash": "101412e7166dcfecb21581b7cc706b80c1e95de805f8ec799ba77005b6cee547", + "component_budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 31, + "serialized_chars": 26681, + "estimated_input_tokens": 6671, + "estimated_eval_tokens": 13598 + }, + "budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 31, + "serialized_chars": 26681, + "estimated_input_tokens": 6671, + "estimated_eval_tokens": 13598 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are mostly unrelated workflow-maintenance PRs. The only dense area is pr-ci-security-gate.yml, but those PRs make different follow-up edits rather than the same concrete fix, so I would not deduplicate any pair.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No true duplicate PR stands out. If a representative is needed, PR 46554 is the closest because it makes the main functional security-gate logic change; the others are smaller or distinct follow-ups.", + "best_issue_reason": null, + "best_pr_reason": "PR 46554 is the strongest standalone PR: it adds actual security-gate behavior (checkout and changed-file checks), whereas the others are narrower workflow tweaks, permissions changes, or comments.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46138", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and different fixes; same broad CI area only." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46558", + "accept": false, + "reason": "Unrelated workflow update versus security-gate permissions tweak." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46563", + "accept": false, + "reason": "Command input handling vs. explanatory note in another workflow." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46716", + "accept": false, + "reason": "Different workflow changes; no shared concrete bug." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46504", + "accept": false, + "reason": "Separate shell-quoting fixes in different workflows." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46503", + "accept": false, + "reason": "SSH runner hardening vs. benchmark trigger logic; distinct changes." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46501", + "accept": false, + "reason": "Different workflows and different problems." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46502", + "accept": false, + "reason": "Unrelated workflow edits; no same code-path issue." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46554", + "accept": false, + "reason": "Security hardening in SSH runner vs. PR-CI gate logic." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46558", + "accept": false, + "reason": "Different workflow file and different fix scope." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46563", + "accept": false, + "reason": "No shared underlying bug; one is SSH hardening, the other is a note." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46716", + "accept": false, + "reason": "Distinct SSH workflow vs. PR CI security gate change." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46503", + "accept": false, + "reason": "Different workflows and triggers; not the same change." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46502", + "accept": false, + "reason": "Separate CI workflows; no common concrete bug." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46504", + "accept": false, + "reason": "Permissions move in anti-slop vs. shell input quoting fix." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow logic and distinct intent." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46503", + "accept": false, + "reason": "Base-ref checkout fix vs. benchmark schedule trigger." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46504", + "accept": false, + "reason": "Different workflows and separate fixes." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46554", + "accept": false, + "reason": "Assign-reviewers checkout tweak is unrelated to PR-CI gate logic." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46504", + "accept": false, + "reason": "Benchmark dispatch support vs. docker-image command quoting." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46554", + "accept": false, + "reason": "Benchmark trigger update is unrelated to security-gate checks." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46554", + "accept": false, + "reason": "Same CI subsystem, but different workflows and different bugs." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46558", + "accept": false, + "reason": "Shell quoting fix vs. security-gate permissions/condition change." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow files and change purposes." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46716", + "accept": false, + "reason": "Docker-image workflow fix vs. PR-CI secrets handling." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46558", + "accept": false, + "reason": "Same workflow file, but different follow-up changes, not one bug." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46563", + "accept": false, + "reason": "Actual gate logic vs. a documentation comment." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46716", + "accept": false, + "reason": "Different security-gate aspects: file checks vs. secrets passing." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46563", + "accept": false, + "reason": "Permissions/flow control change versus comment-only update." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46716", + "accept": false, + "reason": "Different fixes to the same file, but not the same concrete problem." + }, + { + "left": "pull_request:46563", + "right": "pull_request:46716", + "accept": false, + "reason": "Comment-only clarification vs. secrets behavior change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 24, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46174-2", + "nodes": [ + "pull_request:46174", + "pull_request:46780" + ], + "soft_pairs": [ + "pull_request:46174|pull_request:46780" + ], + "prepared_review_unit_hash": "a48ce4ca162f555682f79411d7229ac1ccc08d7822dfab654bb881639fccddbc", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6030, + "estimated_input_tokens": 1508, + "estimated_eval_tokens": 3272 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6030, + "estimated_input_tokens": 1508, + "estimated_eval_tokens": 3272 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two PRs target the same reported MPS crash in the sinusoidal position embedding helper, but they implement different fixes and are not safe to treat as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46174 is the broader, more complete fix for the float64-on-MPS crash: it updates the helper across all affected model files and keeps the internal math on CPU float64 before casting back.", + "best_issue_reason": null, + "best_pr_reason": "PR 46174 is the best representative of the cluster because it is the original, more comprehensive fix and has more discussion/review context than the narrower follow-up.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46174", + "right": "pull_request:46780", + "accept": false, + "reason": "They share the same issue target and helper, but the patches are different solutions: 46174 avoids float64 on MPS by doing the arithmetic on CPU first, while 46780 switches the helper to use dtype for internal arithmetic. Those are alternative implementations, not one mergeable duplicate PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 25, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46310-2", + "nodes": [ + "pull_request:46310", + "pull_request:47162" + ], + "soft_pairs": [ + "pull_request:46310|pull_request:47162" + ], + "prepared_review_unit_hash": "3f601733998f19a9ead4753c4aa69f0772a5f37164126243d89f03074f921d36", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4883, + "estimated_input_tokens": 1221, + "estimated_eval_tokens": 2698 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4883, + "estimated_input_tokens": 1221, + "estimated_eval_tokens": 2698 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests address the same off-by-one bug in the sliding-window right mask for Moonshine Streaming; 47162 is the broader, more complete fix and also includes tests plus a related Gemma4 change with the same mask correction.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 47162 is the better canonical PR because it applies the same right-window off-by-one fix in Moonshine Streaming, adds test coverage, and is the more complete implementation of the bug fix across affected files.", + "best_issue_reason": null, + "best_pr_reason": "PR 47162 is the strongest representative of this bug fix set because it contains the same concrete correction as 46310 plus broader coverage and validation, making it the most useful single artifact to point to.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46310", + "right": "pull_request:47162", + "accept": true, + "reason": "Same underlying off-by-one right-window mask bug and same code path; 47162 includes the same Moonshine Streaming fix, so these are effectively overlapping fixes that could be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 26, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46574-2", + "nodes": [ + "pull_request:46574", + "pull_request:46580" + ], + "soft_pairs": [ + "pull_request:46574|pull_request:46580" + ], + "prepared_review_unit_hash": "72a4846d343e0e9ed90dcdca126d61b306fd0a1c68f721e6db7034705f38b72d", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4409, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4409, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are both tokenizer-related, but they fix different bugs in different code paths: CodeLlama decode stripping vs. DeepSeek model-class selection/loading. They should not be clustered as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR for this cluster: the PRs address distinct tokenizer defects and are not mergeable into one change without conflating unrelated fixes.", + "best_issue_reason": null, + "best_pr_reason": "PR #46574 is the better representative of a concrete, self-contained bug fix with a clear failure mode and explicit issue link, while #46580 is a separate checkpoint-routing workaround.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46574", + "right": "pull_request:46580", + "accept": false, + "reason": "Different underlying problems and code paths: #46574 fixes CodeLlama decode behavior removing real leading spaces, while #46580 changes AutoTokenizer checkpoint-prefix routing for DeepSeek models loading the wrong tokenizer class." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 27, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47122-13", + "nodes": [ + "pull_request:47122", + "pull_request:47123", + "pull_request:47124", + "pull_request:47224", + "pull_request:47266", + "pull_request:47280", + "pull_request:47281", + "pull_request:47282", + "pull_request:47310", + "pull_request:47380", + "pull_request:47443", + "pull_request:47444", + "pull_request:47445" + ], + "soft_pairs": [ + "pull_request:47122|pull_request:47281", + "pull_request:47122|pull_request:47123", + "pull_request:47122|pull_request:47280", + "pull_request:47122|pull_request:47224", + "pull_request:47122|pull_request:47310", + "pull_request:47123|pull_request:47280", + "pull_request:47123|pull_request:47443", + "pull_request:47123|pull_request:47266", + "pull_request:47123|pull_request:47281", + "pull_request:47124|pull_request:47310", + "pull_request:47124|pull_request:47445", + "pull_request:47124|pull_request:47282", + "pull_request:47124|pull_request:47380", + "pull_request:47124|pull_request:47443", + "pull_request:47266|pull_request:47443", + "pull_request:47266|pull_request:47280", + "pull_request:47266|pull_request:47444", + "pull_request:47266|pull_request:47445", + "pull_request:47280|pull_request:47443", + "pull_request:47280|pull_request:47281", + "pull_request:47280|pull_request:47310", + "pull_request:47281|pull_request:47443", + "pull_request:47281|pull_request:47445", + "pull_request:47282|pull_request:47380", + "pull_request:47282|pull_request:47310", + "pull_request:47280|pull_request:47282", + "pull_request:47282|pull_request:47444", + "pull_request:47310|pull_request:47380", + "pull_request:47310|pull_request:47445", + "pull_request:47280|pull_request:47380", + "pull_request:47380|pull_request:47444", + "pull_request:47443|pull_request:47445", + "pull_request:47443|pull_request:47444", + "pull_request:47444|pull_request:47445", + "pull_request:47280|pull_request:47444", + "pull_request:47123|pull_request:47444", + "pull_request:47282|pull_request:47445" + ], + "prepared_review_unit_hash": "0e8e629c20e3ca76318adb85e414727fac470f4415fa350733a7ec43a9d1521c", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 37, + "serialized_chars": 44721, + "estimated_input_tokens": 11181, + "estimated_eval_tokens": 22618 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 37, + "serialized_chars": 44721, + "estimated_input_tokens": 11181, + "estimated_eval_tokens": 22618 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly false-positive similarity across unrelated serge-generated test-fix PRs. True duplicates only appear within the MiniMax-M3-VL load_error pair, the Mamba2 OOM pair, and the DeepSeek rope/Yarn cleanup pair.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 47443 is the clearest representative of the shared DeepSeek/MoE rope/Yarn code cleanup that also overlaps 47444; it has the strongest common code-path change among the PRs.", + "best_issue_reason": null, + "best_pr_reason": "PR 47443 is the best global representative because it carries the broad shared model-side fix across the DeepSeek/MoE files, while the nearby PRs are either narrower duplicates or mixed with unrelated changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47122", + "right": "pull_request:47281", + "accept": false, + "reason": "Different models and unrelated fixes; only the automation/template looks similar." + }, + { + "left": "pull_request:47122", + "right": "pull_request:47123", + "accept": false, + "reason": "Edgetam import_or_config bug vs MiniMax-M3-VL load_error; not the same change." + }, + { + "left": "pull_request:47122", + "right": "pull_request:47280", + "accept": false, + "reason": "Different model and failure; same test-fix pattern but not the same underlying bug." + }, + { + "left": "pull_request:47122", + "right": "pull_request:47224", + "accept": false, + "reason": "47224 is a bundled multi-fix PR; the shared Edgetam hunk is not enough to treat them as the same PR." + }, + { + "left": "pull_request:47122", + "right": "pull_request:47310", + "accept": false, + "reason": "Edgetam vs Gemma3; different code paths and failure modes." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47266", + "accept": false, + "reason": "MiniMax-M3-VL load_error vs DeepSeek-V3.2 load_error; same symptom, different model and fix." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47280", + "accept": true, + "reason": "Same MiniMax-M3-VL load_error; both add temporary offload-folder handling in the same loader/test path." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47281", + "accept": false, + "reason": "MiniMax-M3-VL load_error vs glm46v/glm4v modeling change; not the same bug." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47443", + "accept": false, + "reason": "MiniMax test load_error and DeepSeek rope cleanup are unrelated fixes." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47444", + "accept": false, + "reason": "Different models and different fixes; no shared concrete code-path problem." + }, + { + "left": "pull_request:47124", + "right": "pull_request:47310", + "accept": false, + "reason": "Falcon-Mamba output mismatch vs Gemma3 output mismatch; same symptom category only." + }, + { + "left": "pull_request:47124", + "right": "pull_request:47445", + "accept": false, + "reason": "Falcon-Mamba generation test vs PEFT loading fix; unrelated code paths." + }, + { + "left": "pull_request:47124", + "right": "pull_request:47282", + "accept": false, + "reason": "Falcon-Mamba output mismatch vs Mamba2 OOM; different underlying issues." + }, + { + "left": "pull_request:47124", + "right": "pull_request:47380", + "accept": false, + "reason": "Falcon-Mamba output mismatch vs Mamba2 OOM; only broad test-fix similarity." + }, + { + "left": "pull_request:47124", + "right": "pull_request:47443", + "accept": false, + "reason": "Falcon-Mamba generation output issue is unrelated to the DeepSeek rope cleanup." + }, + { + "left": "pull_request:47266", + "right": "pull_request:47443", + "accept": false, + "reason": "DeepSeek-V3.2 load_error test workaround vs model implementation cleanup; not the same concrete bug." + }, + { + "left": "pull_request:47266", + "right": "pull_request:47280", + "accept": false, + "reason": "Different models and different load_error workarounds; no shared code-path." + }, + { + "left": "pull_request:47266", + "right": "pull_request:47444", + "accept": false, + "reason": "DeepSeek-V3.2 test load_error vs MPT model cleanup; unrelated." + }, + { + "left": "pull_request:47266", + "right": "pull_request:47445", + "accept": false, + "reason": "DeepSeek-V3.2 load_error test fix vs PEFT integration fix; unrelated changes." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47281", + "accept": false, + "reason": "MiniMax-M3-VL load_error vs glm46v/glm4v refactor; different models and bugs." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47310", + "accept": false, + "reason": "MiniMax-M3-VL load_error vs Gemma3 output mismatch; not the same issue." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47380", + "accept": false, + "reason": "Same broad OOM theme, but different model/test and different mitigation; not a duplicate." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47443", + "accept": false, + "reason": "MiniMax loader workaround vs DeepSeek rope cleanup; unrelated." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47444", + "accept": false, + "reason": "MiniMax loader workaround vs MPT/DeepSeek code cleanup; not the same bug." + }, + { + "left": "pull_request:47281", + "right": "pull_request:47443", + "accept": false, + "reason": "glm46v/glm4v model refactor vs DeepSeek rope cleanup; unrelated." + }, + { + "left": "pull_request:47281", + "right": "pull_request:47445", + "accept": false, + "reason": "glm46v/glm4v refactor vs PEFT integration change; different code paths." + }, + { + "left": "pull_request:47282", + "right": "pull_request:47380", + "accept": true, + "reason": "Same Mamba2 OOM in the same integration test; both adjust memory/runtime behavior to avoid the crash." + }, + { + "left": "pull_request:47282", + "right": "pull_request:47310", + "accept": false, + "reason": "Mamba2 OOM vs Gemma3 output mismatch; unrelated failures." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47282", + "accept": false, + "reason": "MiniMax-M3-VL load_error and Mamba2 OOM are different model failures." + }, + { + "left": "pull_request:47282", + "right": "pull_request:47444", + "accept": false, + "reason": "Mamba2 OOM workaround vs MPT/DeepSeek cleanup; not the same change." + }, + { + "left": "pull_request:47310", + "right": "pull_request:47380", + "accept": false, + "reason": "Gemma3 output mismatch vs Mamba2 OOM; different underlying bugs." + }, + { + "left": "pull_request:47310", + "right": "pull_request:47445", + "accept": false, + "reason": "Gemma3 generation/default-cache fix vs PEFT loading fix; unrelated." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47380", + "accept": false, + "reason": "MiniMax-M3-VL load_error and Mamba2 OOM are different model-specific issues." + }, + { + "left": "pull_request:47380", + "right": "pull_request:47444", + "accept": false, + "reason": "Mamba2 OOM mitigation vs DeepSeek/MPT code cleanup; unrelated." + }, + { + "left": "pull_request:47443", + "right": "pull_request:47445", + "accept": false, + "reason": "47445 adds a PEFT-specific load fix in addition to overlapping model-file edits; not the same concrete change." + }, + { + "left": "pull_request:47443", + "right": "pull_request:47444", + "accept": true, + "reason": "They share the same DeepSeek/MoE rope/Yarn cleanup across the same model files; only the targeted failing test differs." + }, + { + "left": "pull_request:47444", + "right": "pull_request:47445", + "accept": false, + "reason": "Shared model-file overlap, but 47445 is mixed with an unrelated PEFT integration fix." + }, + { + "left": "pull_request:47280", + "right": "pull_request:47444", + "accept": false, + "reason": "MiniMax-M3-VL loader workaround vs DeepSeek/MPT code cleanup; not the same bug." + }, + { + "left": "pull_request:47123", + "right": "pull_request:47444", + "accept": false, + "reason": "MiniMax-M3-VL load_error workaround vs MPT/DeepSeek cleanup; different underlying issue." + }, + { + "left": "pull_request:47282", + "right": "pull_request:47445", + "accept": false, + "reason": "Mamba2 OOM workaround vs PEFT integration and model-file cleanup; not the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 28, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47140-2", + "nodes": [ + "pull_request:47140", + "pull_request:47303" + ], + "soft_pairs": [ + "pull_request:47140|pull_request:47303" + ], + "prepared_review_unit_hash": "189292c1154dd9dfd6e9871e379a602e14017e8fb9bb897b5cefada8431c7e32", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4409, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4409, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same chat CLI bug where management endpoints are built from a `/v1` base URL instead of the service root. PR #47140 is the broader fix; PR #47303 is a narrower variant for the same path.", + "confidence": 0.82, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #47140 is the better canonical representative because it fixes the base URL normalization once in `RichInterface`, covering management calls consistently rather than patching only one call site.", + "best_issue_reason": null, + "best_pr_reason": "PR #47140 is the strongest overall candidate: it directly addresses the `/v1` base handling at initialization and appears more complete for the underlying routing bug.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47140", + "right": "pull_request:47303", + "accept": true, + "reason": "Same underlying bug: chat management URLs are incorrectly rooted at `/v1`. Both PRs fix that service-root routing in `src/transformers/cli/chat.py`, and they could be consolidated into one change by choosing one implementation approach." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs target chat CLI management URL routing away from a `/v1` base, and the distinction between a broader centralized fix (#47140) and a narrower call-site fix (#47303) is supported by the diffs and titles. The soft-edge mergeability verdict is conservative enough: they appear to address the same bug and could be combined into one PR by selecting one implementation approach rather than both." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 29, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47261-2", + "nodes": [ + "pull_request:47261", + "pull_request:47262" + ], + "soft_pairs": [ + "pull_request:47261|pull_request:47262" + ], + "prepared_review_unit_hash": "7a46b43c3f55ce21663956bacbe3ba98b57dfe48acd1bfb1d0d3dd0839c7276f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4419, + "estimated_input_tokens": 1105, + "estimated_eval_tokens": 2466 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4419, + "estimated_input_tokens": 1105, + "estimated_eval_tokens": 2466 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster contains two unrelated open PRs: one fixes gradient inflation with label smoothing during gradient accumulation, and the other adds/validates `min_tokens_to_keep` handling in `TopHLogitsWarper`. They target different code paths and are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 47261 is the stronger canonical representative because it addresses a concrete training bug, has an explicit issue link, and its diff spans the actual fix plus tests; the other PR is a separate generation feature/fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 47261 is the best overall representative of the cluster since it is more clearly bug-driven and self-contained, while PR 47262 is unrelated in subsystem and behavior.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47261", + "right": "pull_request:47262", + "accept": false, + "reason": "Different subsystems and behaviors: one changes Trainer/LabelSmoother loss scaling for gradient accumulation, the other changes generation logits warping to respect `min_tokens_to_keep`. They do not fix the same underlying bug and would not plausibly merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 30, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47315-2", + "nodes": [ + "pull_request:47315", + "pull_request:47430" + ], + "soft_pairs": [ + "pull_request:47315|pull_request:47430" + ], + "prepared_review_unit_hash": "8858cd1794ea2e0b7a8ea369a7e165f385b808df7437418d147d8559f3da0795", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4383, + "estimated_input_tokens": 1096, + "estimated_eval_tokens": 2448 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4383, + "estimated_input_tokens": 1096, + "estimated_eval_tokens": 2448 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs touch compressed_tensors MoE expert decompression, but they address different bugs: one fixes fallback shape reconstruction for non-32-divisible bit widths, the other preserves asymmetric zero_point handling. Related subsystem, not the same change.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 47430 is the broader decompression fix: it updates both the integration path and the quantizer to carry zero_point through asymmetric MoE expert decompression, whereas PR 47315 only patches shape inference for certain bit widths.", + "best_issue_reason": null, + "best_pr_reason": "PR 47430 is the better representative of the cluster because it covers a more complete decompression-path change across multiple files, while PR 47315 is a narrower single-file bug fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47315", + "right": "pull_request:47430", + "accept": false, + "reason": "Different concrete defects: 47315 fixes incorrect unpacked-dimension fallback for 3/5/6/7-bit packing, while 47430 fixes missing zero_point propagation for asymmetric decompression. Same area, not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 31, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47400-2", + "nodes": [ + "pull_request:47400", + "pull_request:47421" + ], + "soft_pairs": [ + "pull_request:47400|pull_request:47421" + ], + "prepared_review_unit_hash": "ea1b59c27dbb00398b91a2f10f7e4e2137e4fe5d17f26d75cdaf52a2d433d604", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4619, + "estimated_input_tokens": 1155, + "estimated_eval_tokens": 2566 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4619, + "estimated_input_tokens": 1155, + "estimated_eval_tokens": 2566 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs touch the same GGUF loading/documentation/test area, but they add support for different model families: Phi-2 vs DeepSeek-V3. They are related by subsystem, not by the same underlying change.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #47421 is the broader representative of the cluster: it adds GGUF support for DeepSeek-V3 with a larger code path update and docs/test changes, making it a better anchor than the narrower Phi-2-only PR.", + "best_issue_reason": null, + "best_pr_reason": "PR #47421 is the best PR candidate because it appears to be the more substantial and more representative GGUF integration change, while #47400 is a separate model-specific addition.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47400", + "right": "pull_request:47421", + "accept": false, + "reason": "Both PRs modify the same GGUF integration files, but they target different model architectures (Phi-2 vs DeepSeek-V3) and are not the same concrete bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 32, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47431-2", + "nodes": [ + "pull_request:47431", + "pull_request:47434" + ], + "soft_pairs": [ + "pull_request:47431|pull_request:47434" + ], + "prepared_review_unit_hash": "13cba77cceee4848611f83b8a4724d5c5199e766bfc5c2744c9be77c7bb95037", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3811, + "estimated_input_tokens": 953, + "estimated_eval_tokens": 2162 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3811, + "estimated_input_tokens": 953, + "estimated_eval_tokens": 2162 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests make the same tokenizers upper-bound relaxation in the same two files and target the same issue, so they are duplicates of one another.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #47434 is the better canonical PR because its title is more explicit and it directly references the tracked issue (#47429), while the diff is identical to #47431.", + "best_issue_reason": null, + "best_pr_reason": "PR #47434 is the strongest representative of the cluster: it states the exact change, links the issue, and has the same code edits as #47431.", + "soft_edge_verdicts": [ + { + "left": "pull_request:47431", + "right": "pull_request:47434", + "accept": true, + "reason": "Identical diff, same touched files, and the same explicit issue target; these are the same underlying change and could be merged as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + } + ] +} diff --git a/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/manifest.json b/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..37f8c4905fabc4384c87fe8014055341984e20d8 --- /dev/null +++ b/snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260721t060018z", + "artifacts": { + "hybrid": "snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-07-21T06:15:11Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260721T060018Z", + "variant": "hybrid" +} diff --git a/snapshots/20260721T060018Z/manifest.json b/snapshots/20260721T060018Z/manifest.json index c3cb3380cb9f2015d6c06adb38c184ea8e3bc4b4..66909e9f09b76117ca6f254ede0539507fc0a063 100644 --- a/snapshots/20260721T060018Z/manifest.json +++ b/snapshots/20260721T060018Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 989 }, "extracted_at": "2026-07-21T06:00:18Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260721t060018z", + "runs": { + "hybrid-model-20260721t060018z": { + "analysis_id": "hybrid-model-20260721t060018z", + "artifacts": { + "hybrid": "snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260721T060018Z/analysis-runs/hybrid-model-20260721t060018z/manifest.json", + "model": null, + "published_at": "2026-07-21T06:15:11Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260721T060018Z", "watermark": {