diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index 417cc23785f55a26620980cb9ad8c83747dcf670..0b2f27ac7682a9cbd282f079e2e26198adf47d3f 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": "20260422T060051Z", - "generated_at": "2026-04-22T06:09:46Z", + "snapshot_id": "20260422T120052Z", + "generated_at": "2026-04-22T12:09:12Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ @@ -1053,7 +1053,7 @@ "cluster_id": "cluster-41211-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #41211.", "status": "open", - "confidence": 0.55, + "confidence": 0.75, "canonical_issue_number": 41211, "canonical_pr_number": 44339, "issue_numbers": [ @@ -1064,15 +1064,16 @@ 44339 ], "evidence_types": [ - "closing_reference" + "closing_reference", + "shared_issue_target" ], "pr_comparisons": [ { "left_pr_number": 41356, "right_pr_number": 44339, - "code_similarity": 0.164, - "size_similarity": 0.095, - "file_overlap": 0.3, + "code_similarity": 0.156, + "size_similarity": 0.094, + "file_overlap": 0.273, "area_overlap": 0.078, "patch_similarity": 0.183, "shared_filenames": [ @@ -1347,212 +1348,48 @@ ] }, { - "cluster_id": "cluster-43366-4", - "summary": "Cluster of 1 issues and 3 PRs centered on issue #43366.", + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", "status": "open", - "confidence": 0.8, - "canonical_issue_number": 43366, - "canonical_pr_number": 45506, + "confidence": 0.75, + "canonical_issue_number": 43240, + "canonical_pr_number": 43251, "issue_numbers": [ - 43366 + 43240 ], "pr_numbers": [ - 43757, - 45500, - 45506 + 43251, + 43254 ], "evidence_types": [ "closing_reference", - "shared_issue_target", - "soft_similarity" + "shared_issue_target" ], "pr_comparisons": [ { - "left_pr_number": 43757, - "right_pr_number": 45500, - "code_similarity": 0.096, - "size_similarity": 0.149, - "file_overlap": 0.2, - "area_overlap": 0.005, - "patch_similarity": 0.031, - "shared_filenames": [ - "src/transformers/modeling_gguf_pytorch_utils.py" - ], - "shared_file_areas": [ - { - "filename": "src/transformers/modeling_gguf_pytorch_utils.py", - "left_ranges": [ - [ - 462, - 474 - ] - ], - "right_ranges": [ - [ - 171, - 277 - ], - [ - 456, - 462 - ], - [ - 518, - 525 - ], - [ - 620, - 627 - ], - [ - 709, - 753 - ] - ] - } - ] - }, - { - "left_pr_number": 43757, - "right_pr_number": 45506, - "code_similarity": 0.116, - "size_similarity": 0.17, - "file_overlap": 0.25, - "area_overlap": 0.005, - "patch_similarity": 0.034, - "shared_filenames": [ - "src/transformers/modeling_gguf_pytorch_utils.py" - ], - "shared_file_areas": [ - { - "filename": "src/transformers/modeling_gguf_pytorch_utils.py", - "left_ranges": [ - [ - 462, - 474 - ] - ], - "right_ranges": [ - [ - 171, - 277 - ], - [ - 456, - 462 - ], - [ - 518, - 525 - ], - [ - 620, - 627 - ], - [ - 709, - 753 - ] - ] - } - ] - }, - { - "left_pr_number": 45500, - "right_pr_number": 45506, - "code_similarity": 0.886, - "size_similarity": 0.877, - "file_overlap": 0.75, - "area_overlap": 0.996, - "patch_similarity": 0.912, + "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/integrations/ggml.py", - "src/transformers/modeling_gguf_pytorch_utils.py", - "tests/quantization/ggml/test_ggml.py" + "src/transformers/loss/loss_utils.py" ], "shared_file_areas": [ { - "filename": "src/transformers/integrations/ggml.py", - "left_ranges": [ - [ - 89, - 109 - ] - ], - "right_ranges": [ - [ - 89, - 109 - ] - ] - }, - { - "filename": "src/transformers/modeling_gguf_pytorch_utils.py", - "left_ranges": [ - [ - 171, - 277 - ], - [ - 456, - 462 - ], - [ - 518, - 525 - ], - [ - 620, - 627 - ], - [ - 709, - 753 - ] - ], - "right_ranges": [ - [ - 171, - 277 - ], - [ - 456, - 462 - ], - [ - 518, - 525 - ], - [ - 620, - 627 - ], - [ - 709, - 753 - ] - ] - }, - { - "filename": "tests/quantization/ggml/test_ggml.py", + "filename": "src/transformers/loss/loss_utils.py", "left_ranges": [ [ - 351, - 358 - ], - [ - 386, - 406 + 30, + 50 ] ], "right_ranges": [ [ - 351, - 358 - ], - [ - 386, - 405 + 30, + 43 ] ] } @@ -1561,15 +1398,14 @@ ] }, { - "cluster_id": "cluster-45081-4", - "summary": "Cluster of 2 issues and 2 PRs centered on issue #45081.", + "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, - 45356 + 45081 ], "pr_numbers": [ 45086, @@ -1620,56 +1456,6 @@ } ] }, - { - "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-43698-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", @@ -1729,273 +1515,49 @@ ] }, { - "cluster_id": "cluster-44018-2", - "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "cluster_id": "cluster-45563-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45563.", "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 44068, - "issue_numbers": [], + "confidence": 0.8, + "canonical_issue_number": 45563, + "canonical_pr_number": 45565, + "issue_numbers": [ + 45563 + ], "pr_numbers": [ - 44018, - 44068 + 45565, + 45575 ], "evidence_types": [ + "closing_reference", + "shared_issue_target", "soft_similarity" ], "pr_comparisons": [ { - "left_pr_number": 44018, - "right_pr_number": 44068, - "code_similarity": 0.766, - "size_similarity": 0.939, + "left_pr_number": 45565, + "right_pr_number": 45575, + "code_similarity": 0.709, + "size_similarity": 0.667, "file_overlap": 1.0, - "area_overlap": 0.425, - "patch_similarity": 0.866, + "area_overlap": 0.526, + "patch_similarity": 0.607, "shared_filenames": [ - "src/transformers/models/gpt_neo/modeling_gpt_neo.py", - "tests/models/gpt_neo/test_modeling_gpt_neo.py" + "src/transformers/generation/utils.py" ], "shared_file_areas": [ { - "filename": "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "filename": "src/transformers/generation/utils.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 + 2300, + 2318 ] ], "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 + 2305, + 2314 ] ] } @@ -2004,16 +1566,7 @@ ] } ], - "duplicate_issues": [ - { - "cluster_id": "cluster-45081-4", - "canonical_issue_number": 45081, - "duplicate_issue_numbers": [ - 45356 - ], - "reason": "Issues in cluster-45081-4 are treated as duplicates because they share closing_reference, shared_issue_target, soft_similarity evidence." - } - ], + "duplicate_issues": [], "duplicate_prs": [ { "cluster_id": "cluster-41211-3", @@ -2022,7 +1575,7 @@ 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." + "reason": "PRs in cluster-41211-3 are treated as duplicates because they converge on issue #41211 with closing_reference, shared_issue_target evidence." }, { "cluster_id": "cluster-43240-3", @@ -2033,16 +1586,6 @@ "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-43366-4", - "canonical_pr_number": 45506, - "duplicate_pr_numbers": [ - 43757, - 45500 - ], - "target_issue_number": 43366, - "reason": "PRs in cluster-43366-4 are treated as duplicates because they converge on issue #43366 with closing_reference, shared_issue_target, soft_similarity evidence." - }, { "cluster_id": "cluster-43656-4", "canonical_pr_number": 43656, @@ -2080,29 +1623,29 @@ "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, + "cluster_id": "cluster-45081-3", + "canonical_pr_number": 45317, "duplicate_pr_numbers": [ - 44018 + 45086 ], - "target_issue_number": null, - "reason": "PRs in cluster-44018-2 are treated as duplicates because they share soft_similarity evidence." + "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-45081-4", - "canonical_pr_number": 45317, + "cluster_id": "cluster-45563-3", + "canonical_pr_number": 45565, "duplicate_pr_numbers": [ - 45086 + 45575 ], - "target_issue_number": 45081, - "reason": "PRs in cluster-45081-4 are treated as duplicates because they converge on issue #45081 with closing_reference, shared_issue_target, soft_similarity evidence." + "target_issue_number": 45563, + "reason": "PRs in cluster-45563-3 are treated as duplicates because they converge on issue #45563 with closing_reference, shared_issue_target, soft_similarity evidence." } ], "best_issue": { - "cluster_id": "cluster-43366-4", - "issue_number": 43366, - "reason": "Issue #43366 is the strongest global issue candidate because it is open, belongs to a cluster with 4 artifacts, and carries 5 discussion comments plus 7 inbound references.", - "score": 157.0 + "cluster_id": "cluster-41211-3", + "issue_number": 41211, + "reason": "Issue #41211 is the strongest global issue candidate because it is open, belongs to a cluster with 3 artifacts, and carries 6 discussion comments plus 6 inbound references.", + "score": 148.0 }, "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 753a3e5fc31436eb0a3fdc0d595f128737f576ca..0c65014f0cf6cc4bf5af227a6d29f8f07c19a813 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": "20260422T060051Z", - "generated_at": "2026-04-22T06:09:46Z", + "snapshot_id": "20260422T120052Z", + "generated_at": "2026-04-22T12:09:12Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -26,7 +26,7 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "310768fbd0b727863706c29c0e02b2d2d6c0c6f69e83a841e3b7d62dbb74c7f4", + "prepared_review_unit_hash": "cf649f9b9856122366cdb67fd4649209142ca58ef11b06e55c7b69ac3efe0c11", "component_budget": { "node_count": 3, "item_count": 3, @@ -48,24 +48,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The three issues do not look like the same underlying bug. #16618 is a general model-loading failure, #15980 is specifically about a bad authentication/error message when downloading a private model, and #13244 is about TAPAS tokenization mismatch with TensorFlow code. I would keep them separate.", - "confidence": 0.82, - "canonical_issue_reason": "#16618 is the broadest and most central of the three, focused on model loading, so it is the best representative issue for this cluster even though the others are not true duplicates.", + "summary": "The cluster does not look like a single duplicate set. One issue is unrelated (Tapas tokenization), and the two model-loading issues are only loosely similar from their titles alone.", + "confidence": 0.73, + "canonical_issue_reason": "Issue #16618 is the broadest and most general model-loading report in the group, so it is the best candidate to serve as the hub if these were to be grouped.", "canonical_pr_reason": null, - "best_issue_reason": "#16618 is the strongest standalone issue because it states a clear user-facing failure and is the most generally applicable of the set.", + "best_issue_reason": "Issue #16618 is the best representative because its title is generic and could subsume related loading failures, while #15980 is specifically about private-model authentication and #13244 is unrelated.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Tokenization behavior in TAPAS is a different subsystem and bug class from a model-loading failure; no evidence they share the same root cause." + "reason": "Different underlying problems: TAPAS tokenization mismatch versus a model-loading failure for bert-base-uncased. Same broad ML ecosystem, but not the same bug." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "#15980 is about authentication/download messaging for private models, while #16618 is a general inability to load 'bert-base-uncased'; these are not the same concrete issue." + "reason": "#15980 is specifically about a bad auth/error message when downloading a private model, while #16618 concerns failing to load bert-base-uncased. Titles suggest different failure modes and likely different fixes." } ] }, @@ -203,316 +203,370 @@ "issue:41093|issue:41762", "issue:43296|issue:43366" ], - "prepared_review_unit_hash": "6d917a365e2efd2146e301b8301b8142d635868e7d01c1d4fc52441c1a3bd136", + "prepared_review_unit_hash": "afe20e379c359238cd511977591dd626b5ca5622b4c7d549505e2f64326a9b49", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 63, - "serialized_chars": 46485, - "estimated_input_tokens": 11622, - "estimated_eval_tokens": 23500 + "serialized_chars": 46482, + "estimated_input_tokens": 11621, + "estimated_eval_tokens": 23498 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous model/runtime/doc issues. The only clear duplicate-looking pair is the two DeepSpeed ZeRO-3 Bert index-out-of-bounds reports; the rest differ by model, symptom, or code path.", + "summary": "This cluster is mostly heterogeneous; only the DeepSpeed ZeRO-3 index-0 failures look like true duplicates. The rest are different model/config/documentation issues with similar wording but distinct root causes.", "confidence": 0.79, - "canonical_issue_reason": "issue:43596 is the cleanest representative of the one true duplicate pair: it is a concrete DeepSpeed ZeRO-3/Bert index-out-of-bounds report and is closely mirrored by issue:43638.", + "canonical_issue_reason": "43638 is the best representative of the repeated DeepSpeed ZeRO-3 index-0-out-of-bounds bug; it most closely matches 43596 and is generic enough to cover the same failure mode.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 is the best global issue representative because it describes the shared ZeRO-3/Bert loading failure most concretely and has the strongest duplicate match.", + "best_issue_reason": "43638 has the clearest duplicate signal and the most reusable description for the shared ZeRO-3 index-0 bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Same family (Qwen3.5) but different failures: missing config field vs streaming chat/completions behavior." + "reason": "Different failures: Qwen3.5 config attribute error vs streaming chat/completions runtime failure." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different models and different failure modes; one is dtype mismatch in a processor path, the other is AMD GPU runtime support." + "reason": "Different models and symptoms; dtype mismatch on Mistral-3 preprocessing is not the same as GPT-OSS failing on AMD GPUs." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Both mention inference/runtime issues, but the underlying bug paths are unrelated (autocast dtype mismatch vs AMD GPU failure)." + "reason": "Both are model/runtime problems, but the autocast dtype mismatch and AMD GPU failure are distinct bugs." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Different models and code paths; one is GPU backend compatibility, the other is an invalid method call in a MoE model." + "reason": "Different code paths: AMD GPU runtime failure vs invalid method call in GraniteMoEHybrid." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Different models and symptoms; NVFP4 runtime error is not the same as the AMD GPU issue." + "reason": "Qwen-3 NVFP4 runtime error is unrelated to the GPT-OSS AMD GPU issue." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "Both are runtime problems, but one is DeepSpeed collective mismatch and the other is ONNX export." + "reason": "DeepSpeed collective mismatch and ONNX export error are unrelated failures." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Unrelated model/export error vs Qwen3Moe FP8 config failure." + "reason": "ModernBERT ONNX export and Qwen3MoE FP8 failure are different model/export paths." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "Different code paths: ONNX export failure vs device-mapping cuda assert." + "reason": "ONNX export vs auto device mapping CUDA assert are not the same bug." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are docs cleanup items, but they refer to different removed features and different pages." + "reason": "Both are v5 docs cleanup issues, but they cover different missing references and not one underlying defect." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "Different bugs: ONNX export failure vs missing GenerationMixin import." + "reason": "Export error and GenerationMixin import error are unrelated." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Both are shape/index errors, but the model/setup and stack traces differ enough that they do not look like the same bug." + "reason": "Similar index/shape wording, but one is ZeRO-3 index-0-out-of-bounds and the other is a mask-vs-tensor shape mismatch." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Different symptom pattern and trigger; one is ZeRO-3 Bert init failure, the other is a mask/tensor shape mismatch." + "reason": "Different concrete errors: ZeRO-3 index-0-out-of-bounds vs mask shape mismatch." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "Unrelated errors in different features: mask shape mismatch vs GPT-OSS tensor-parallel failure." - }, - { - "left": "issue:44805", - "right": "issue:44918", - "accept": false, - "reason": "Different stacks and causes; tensor mask mismatch is not the same as TRL embedding unpacking failure." + "reason": "Mask shape mismatch is unrelated to GPT-OSS tensor-parallel issues." }, { "left": "issue:44661", "right": "issue:44805", "accept": false, - "reason": "Different functionality: tokenizer mapping codegen vs a runtime indexing/shape error." + "reason": "Tokenizer mapping failure is a different bug from the mask/index mismatch error." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "BarkProcessor voice_preset bug is unrelated to ModernBERT ONNX export." + "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 (collective mismatch vs zero-sized index on Bert load)." + "reason": "Both involve DeepSpeed, but one is a collective mismatch and the other is an index-0-out-of-bounds init error." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Same underlying DeepSpeed ZeRO-3/Bert loading bug: both hit index-0-out-of-bounds with zero-sized parameters during init/load." + "reason": "Same DeepSpeed ZeRO-3 index-0-out-of-bounds failure mode on Bert; very likely duplicate reports of the same bug." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Different bugs: detokenization regression vs LoRA merge/save/reload inconsistency." + "reason": "Detokenization bug and LoRA merge/reload inconsistency are different problems." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Different features and errors; Liger cross-entropy dispatch vs GPT-OSS tensor-parallel support." + "reason": "Liger Kernel cross_entropy being called is unrelated to GPT-OSS tensor-parallel loading." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "Unrelated model/runtime error vs LoRA weight merge inconsistency." + "reason": "Qwen-3 NVFP4 runtime error and LoRA save/reload mismatch are separate issues." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Both are tensor-shape-related, but the training stack and specific failure conditions are different." + "reason": "Activation checkpointing/FSDP recomputation mismatch is not the same as the mask shape mismatch bug." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "Both mention DeepSpeed, but the concrete errors and code paths are different." + "reason": "Both mention DeepSpeed, but the training weight-dimension error and collective mismatch are different failure points." }, { "left": "issue:43975", "right": "issue:45237", "accept": false, - "reason": "Unrelated problems: detokenization regression vs AMD GPU compatibility." + "reason": "Different model/runtime paths; detokenization is unrelated to AMD GPU execution failures." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "Different model families and failure modes; ZeRO-3 Bert init bug is unrelated to GPT-OSS tensor parallelism." + "reason": "ZeRO-3 Bert init bug is unrelated to GPT-OSS tensor-parallel support." }, { "left": "issue:34928", "right": "issue:36331", "accept": false, - "reason": "Activation-checkpoint/FSDP tensor recomputation issue is unrelated to the custom trainer signature change." + "reason": "Checkpointing tensor-size mismatch and custom trainer signature mismatch are unrelated." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Different models and subsystems; Bert ZeRO-3 init bug is not the GPT-OSS TP issue." + "reason": "Different bugs and different subsystems." }, { "left": "issue:43716", - "right": "issue:44292", + "right": "issue:44560", "accept": false, - "reason": "Different models and error classes; dtype mismatch in preprocessing vs NVFP4 runtime failure." + "reason": "Mistral-3 dtype mismatch and Qwen3-vl video StopIteration are unrelated." }, { "left": "issue:39290", "right": "issue:45081", "accept": false, - "reason": "Different model/config issues: Gemma3 vLLM attribute error vs Mistral tokenizer regex patch crash." + "reason": "Gemma3 sliding_window_pattern and Mistral tokenizer regex patch are separate code paths." }, { "left": "issue:43828", "right": "issue:45161", "accept": false, - "reason": "Different bug classes: autocast dtype mismatch vs tensor-parallel runtime failure." + "reason": "Autocast dtype mismatch is not the same as GPT-OSS TP failure." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch version incompatibility and Qwen-3 NVFP4 runtime error are unrelated." }, { "left": "issue:44918", "right": "issue:45161", "accept": false, - "reason": "TRl SFT embedding unpacking and GPT-OSS TP failure are unrelated." + "reason": "TRL SFT embedding unpacking failure is different from GPT-OSS tensor-parallel behavior." }, { "left": "issue:34928", "right": "issue:41720", "accept": false, - "reason": "Different training stacks and symptoms; recomputed tensor mismatch vs cuda assert on auto device map." + "reason": "FSDP/activation checkpointing mismatch is unrelated to the CUDA assert on Qwen3 auto device mapping." }, { - "left": "issue:43541", - "right": "issue:43828", + "left": "issue:43827", + "right": "issue:44512", "accept": false, - "reason": "Both are runtime numeric/dtype issues, but they affect different models and code paths (Dynamo grouped_mm vs autocast mismatch)." + "reason": "Both are docs issues about v5 removals, but they target different removed references." }, { - "left": "issue:36010", - "right": "issue:39290", + "left": "issue:43901", + "right": "issue:44512", "accept": false, - "reason": "ImportError for GenerationMixin and Gemma3 config attribute error are unrelated." + "reason": "TextClassificationPipeline docs and run-command docs are distinct documentation problems." }, { - "left": "issue:44841", - "right": "issue:45084", + "left": "issue:39290", + "right": "issue:41720", "accept": false, - "reason": "Processor failure for Voxtral is unrelated to the template compilation error." + "reason": "Gemma3 config attribute error and Qwen3 CUDA assert are unrelated." }, { - "left": "issue:41093", - "right": "issue:41762", + "left": "issue:43866", + "right": "issue:44863", "accept": false, - "reason": "Both are IndexErrors, but one is a mask/tensor length mismatch and the other is ZeRO-3 load-time zero-size indexing." + "reason": "Corrupted Ovis2 checkpoint and NemotronH checkpoint loading failure are different issues." }, { - "left": "issue:43404", + "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "Mistral lm_head tying bug is unrelated to a corrupted Ovis2 checkpoint." + "reason": "Unit test loading failure and corrupted checkpoint are not the same bug." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Both mention Qwen3.5, but one is a chat crash and the other is a streaming inference API failure." + "reason": "Same model family, but different APIs and failures; not the same underlying bug." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper model runtime failure and pydantic PretrainedConfig field breakage are unrelated." + "reason": "Whisper runtime failure and PretrainedConfig pydantic breakage are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Grouped_mm tracing failure and autocast dtype mismatch are different precision/tracing bugs." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "GenerationMixin import error and Gemma3 config attribute error are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure for Voxtral and compile error on non-template nodes are unrelated." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Mistral-3 preprocessing dtype mismatch and GraniteMoEHybrid invalid method are different bugs." + }, + { + "left": "issue:43828", + "right": "issue:44292", + "accept": false, + "reason": "Autocast dtype mismatch and Qwen-3 NVFP4 runtime error are distinct failures." }, { "left": "issue:43039", - "right": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Liger Kernel cross_entropy behavior and TRL SFT embedding unpacking are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": true, + "reason": "Both report the same ZeRO-3 index-0-out-of-bounds failure when loading/training with size-0 tensors; likely the same underlying bug." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs regressions, but they refer to different removed features and separate fixes." + }, + { + "left": "issue:39290", + "right": "issue:41720", "accept": false, - "reason": "Liger kernel dispatch issue and tensor mask mismatch are different bugs." + "reason": "Gemma3 sliding_window_pattern and Qwen3 auto device mapping CUDA assert are unrelated." }, { "left": "issue:43866", "right": "issue:44863", "accept": false, - "reason": "Corrupted checkpoint vs NemotronH load implementation bug are different failure sources." + "reason": "Checkpoint corruption and NemotronH load failure are different issues." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "Unit test model-loading issue and checkpoint corruption are not the same underlying bug." + "reason": "Test loading problem and corrupted checkpoint are not the same root cause." }, { - "left": "issue:43425", + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Different Qwen3.5 scenarios and errors; not one duplicate bug." + }, + { + "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Torch version incompatibility and pydantic model field regression are unrelated." + "reason": "Whisper runtime failure and pydantic model breakage are unrelated." }, { - "left": "issue:39290", - "right": "issue:43531", + "left": "issue:43541", + "right": "issue:43828", "accept": false, - "reason": "Both involve Qwen/Gemma-era config changes, but the actual missing attributes differ." + "reason": "Different precision/tracing failures in different model paths." }, { - "left": "issue:41093", - "right": "issue:43638", + "left": "issue:36010", + "right": "issue:39290", "accept": false, - "reason": "Both are load-time errors, but one is a mask-size mismatch and the other is a zero-size index failure." + "reason": "ImportError and config attribute error are unrelated." }, { - "left": "issue:43296", - "right": "issue:43366", + "left": "issue:44841", + "right": "issue:45084", "accept": false, - "reason": "PaddleOCR-VL load failure and GGUF gpt-oss support request are different issues." + "reason": "Processor failure and template compile error are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative. The accepted duplicate pair, issue:43596 and issue:43638, is strongly supported by the titles and symptom pattern (ZeRO-3, Bert loading, index 0 out of bounds with size 0). The rejected soft-edge pairs are phrased cautiously and do not overclaim similarity." + "feedback": "The summary is grounded in the packet and stays conservative overall. The duplicate calls on 43596/43638 and 41762/43638 are supported by the titles, and the rest of the pairwise rejections generally avoid overclaiming. I didn\u2019t see a soft-edge verdict that clearly overstates the evidence." }, "evaluator_used": true, "retried": false, @@ -618,244 +672,244 @@ "issue:36331|issue:41762", "issue:44560|issue:45507" ], - "prepared_review_unit_hash": "a64926f25a0855480488fc09a04e0a0a48aae5b97bb838b0fe086404c1db2a0f", + "prepared_review_unit_hash": "b2e577c6f3d2140d6913596dd66ccb5e139223286084d688ac13ff9a801f8487", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37565, - "estimated_input_tokens": 9392, - "estimated_eval_tokens": 19040 + "serialized_chars": 37563, + "estimated_input_tokens": 9391, + "estimated_eval_tokens": 19038 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This looks like a heterogeneous set of mostly unrelated issues. The only clear duplicate pair is the non-persistent-buffer regression reports; the rest are just semantically similar within broad subsystems and should stay separate.", - "confidence": 0.88, - "canonical_issue_reason": "Issue 43644 is the best canonical anchor for the only true duplicate family here: it describes the non-persistent-buffer corruption regression broadly and earliest, while later reports like 43950/44534 are narrower restatements.", + "summary": "Most items are unrelated model- or regression-specific issues bundled into one loose cluster. The only clear duplicate subcluster is the non-persistent buffer corruption regression in v5.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43950 is the most specific and diagnostic report of the non-persistent buffer regression, with an explicit v5.x corruption symptom and a clear `from_pretrained()` failure mode.", "canonical_pr_reason": null, - "best_issue_reason": "43644 is the most representative issue in the cluster because it captures the underlying regression most generally; the other items are mostly distinct bugs or feature requests.", + "best_issue_reason": "issue:43950 best represents the only clearly cohesive duplicate set in this cluster; it is the most complete statement of the buffer-corruption regression.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "Different problems: int4 quantization OOM vs tied-weights handling in translation models." + "reason": "Int4 quantization/OOM regression vs tied-weights translation-model issue; different failure modes." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "Both are config-related, but one is missing pad_token_idx and the other is a pydantic PretrainedConfig field regression; not the same bug." + "reason": "Missing `pad_token_idx` in StableLmConfig is separate from the pydantic `PretrainedConfig` field break." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Different MoE/TP failures with different symptoms and code paths; too broad to merge." + "reason": "Torch dynamo grouped_mm tracing crash vs GPT-OSS tensor-parallel support problem; not the same bug." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Autocast dtype mismatch is unrelated to a chat/completions streaming failure." + "reason": "Autocast dtype mismatch on Phi MoE vs Qwen3.5 streaming chat/completions failure." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a different initialization bug than the pydantic field issue." + "reason": "Unexpected `_is_hf_initialized` arg during `init_empty_weights` is distinct from the pydantic config regression." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "GGUF gpt-oss support request is unrelated to Qwen3.5 streaming inference failures." + "reason": "GGUF architecture support request for gpt-oss is not the same as a Qwen3.5 streaming inference bug." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior is unrelated to the pydantic PretrainedConfig regression." + "reason": "Qwen3-MoE `sliding_window` bug is unrelated to `PretrainedConfig`/pydantic breakage." }, { "left": "issue:44661", "right": "issue:45464", "accept": false, - "reason": "Tokenizer-mapping failure for add-new-model-like is a different loading-path issue than streaming inference." + "reason": "Tokenizer-mapping failure in `add-new-model-like` is a different code path from streaming inference." }, { "left": "issue:43054", "right": "issue:43994", "accept": false, - "reason": "Both mention SigLIP2, but one is about text-embedding quality while the other is a model/pipeline result mismatch." + "reason": "Both involve SigLIP2, but one reports degraded embeddings and the other nonsensical AutoModel/pipeline output; too little evidence of one underlying bug." }, { "left": "issue:39290", "right": "issue:43742", "accept": false, - "reason": "Gemma3/vLLM missing sliding_window_pattern and MobileLLM key error are distinct model-loading bugs." + "reason": "Gemma3 `sliding_window_pattern` missing attribute vs MobileLLM key-error on load; different model/config issues." }, { "left": "issue:43541", "right": "issue:44589", "accept": false, - "reason": "Grouped_mm tracing failure and missing Float8 storage object are unrelated runtime errors." + "reason": "Grouped_mm tracing failure and missing Float8 storage are unrelated runtime errors." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Both are IndexError reports, but they involve different shapes, models, and triggers." + "reason": "Mask-shape mismatch in one path vs DeepSpeed ZeRO-3 zero-sized tensor indexing in another." }, { "left": "issue:42915", "right": "issue:43541", "accept": false, - "reason": "Qwen3Moe FP8 failure and Mixtral grouped_mm tracing failure are different model/backend issues." + "reason": "FineGrainedFP8Config failure on Qwen3Moe is not the same as Mixtral grouped_mm tracing failure." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "High perplexity on gpt-oss is not the same as a GLM5 model-specific issue." + "reason": "Perplexity regression on gpt-oss-20b vs a GLM5 issue; no shared underlying bug shown." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP weight shape error and TRL unpacking failure for Qwen3.5 are different code paths." + "reason": "FSDP `'weight' must be 2-D` and TRL embedding unpacking failure are distinct training-time errors." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Untied lm_head weights in Mistral3 is unrelated to translation-model tied-weights issues in v5." + "reason": "Both mention tied weights, but the affected models and reported failures differ too much to merge as one bug." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Jupyter custom model initialization regression is unrelated to flash-attn2 fallback blocking." + "reason": "Notebook custom-model initialization breakage is unrelated to flash-attn2 fallback checks." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash is a different Qwen3.5 cache/backend problem, not tied weights." + "reason": "PagedAttentionCache group-type crash is not the same underlying issue as tied weights." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading failure is unrelated to translation-model tied-weights behavior." + "reason": "NemotronH checkpoint loading failure is unrelated to translation-model tied weights." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization/offloading behavior and tied-weight translation models are not the same underlying bug." + "reason": "Quantization/offloading behavior and tied-weights regression are different problems." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "DSA indexer ReLU discussion is unrelated to docs mentioning the removed run command." + "reason": "A DSA-indexer discussion/joke is not the same as docs mentioning a removed run command." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Qwen3Moe FP8 failure and flash-attn2 fallback gating are different issues." + "reason": "Qwen3Moe FP8 support failure vs flash-attn fallback gating; different fixes." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "TRL unpacking failure for Qwen3.5 is unrelated to GraniteMoEHybrid invalid method calls." + "reason": "TRL SFT embedding unpacking error and GraniteMoEHybrid invalid-method call are unrelated." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Incorrect detokenization in DeepSeek Coder is unrelated to Qwen-3-8B-NVFP4 runtime errors." + "reason": "Detokenization bug in deepseek-coder is separate from the Qwen-3-8B NVFP4 runtime error." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "A typo in cache strategies is not the same as stale docs about a removed run command." + "reason": "Cache-strategy typo and stale docs about a removed command are unrelated documentation issues." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Same underlying regression: non-persistent buffers being corrupted/fill with junk in Transformers 5.x." + "reason": "Both report the same v5 regression where non-persistent buffers are corrupted/filled with junk after loading." }, { "left": "issue:43644", "right": "issue:43950", "accept": true, - "reason": "Clear duplicate reports of the same non-persistent-buffer corruption bug in from_pretrained()/5.x." + "reason": "Same non-persistent-buffer corruption regression; 43950 is the more specific description of the bug." }, { "left": "issue:44292", "right": "issue:45507", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime error and GraniteMoEHybrid invalid method are unrelated." + "reason": "NVFP4 runtime failure and GraniteMoEHybrid invalid-method error do not appear to share a code path." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Autocast dtype mismatch is not the same as the RLHF gradient explosion/position_ids issue." + "reason": "Autocast dtype mismatch on Phi MoE vs RLHF NaN explosion from 3D position_ids and SDPA fallback." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "SigLIP2 docs example errors and SigLIP2 embedding-quality regression are different problems." + "reason": "SigLIP2 docs example problems are different from a report about weak text embeddings." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Bad AutoTokenizer error messaging for Voxtral is unrelated to missing all_tied_weights_keys on Molmo." + "reason": "Bad AutoTokenizer error messaging for Voxtral is separate from missing `all_tied_weights_keys` on MolmoForCausalLM." }, { "left": "issue:43866", "right": "issue:44403", "accept": false, - "reason": "Corrupted checkpoint and generic loading noise are not the same bug." + "reason": "Corrupted Ovis2 checkpoint vs generic loading noise; not the same defect." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "CustomTrainer unexpected kwarg and Gemma3 DeepSpeed ZeRO-3 index error are unrelated." + "reason": "Trainer `compute_loss()` kwarg breakage is unrelated to DeepSpeed ZeRO-3 loading/index errors." }, { "left": "issue:44560", "right": "issue:45507", "accept": false, - "reason": "Qwen3-vl-embedding video StopIteration is unrelated to GraniteMoEHybrid invalid method calls." + "reason": "Qwen3-vl video `StopIteration` and GraniteMoEHybrid invalid-method issues are distinct." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The only accepted duplicate family, 43644/43950/44534, is well supported by the titles, and the rest of the soft-edge rejections are phrased cautiously enough given the limited evidence. No obvious overstatement or mergeability issue stands out." + "feedback": "The analysis is grounded in the packet and stays conservative. The only clearly supported duplicate subcluster is the non-persistent buffer corruption regression (issue:43644 / issue:43950 / issue:44534), and the rest of the pairwise rejections avoid over-merging unlike issues. No obvious evidence overstatement or unmergeable PR-pair claim." }, "evaluator_used": true, "retried": false, @@ -948,8 +1002,8 @@ "issue:44991|issue:45399", "issue:43827|issue:44509", "issue:43792|issue:44220", - "issue:43404|issue:43950", "issue:45081|issue:45399", + "issue:43404|issue:43950", "issue:43257|issue:43866", "issue:42915|issue:45020", "issue:43742|issue:45399", @@ -963,250 +1017,256 @@ "issue:42915|issue:43872", "issue:36331|issue:41093" ], - "prepared_review_unit_hash": "3a9b921dd9bdb92c3373e88b6c5cd8685a879a48c8244a0d418d2ea192b4363e", + "prepared_review_unit_hash": "9955d509fcfbab0e0755a5f6c0f20634de886f6fd6badc5a5c545906cb771698", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 36, - "serialized_chars": 38576, - "estimated_input_tokens": 9644, - "estimated_eval_tokens": 19544 + "serialized_chars": 38581, + "estimated_input_tokens": 9646, + "estimated_eval_tokens": 19548 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most candidates are related only at a broad subsystem level, not the same underlying bug. The only clear duplicate-looking pair is the v5 docs regression about removed pipeline task references.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 44509 is the clearest and broadest statement of the docs regression; 43827 is the narrower variant and reads like the same report.", + "summary": "This cluster is mostly a grab-bag of model-loading and v5 regression reports across unrelated models. Only a few soft edges look like true duplicates: the docs pipeline-removal reports, the SigLIP2 implementation discrepancy pair, and the Whisper fbank/extract feature path pair.", + "confidence": 0.69, + "canonical_issue_reason": "No single perfect canonical issue exists, but issue 45020 is the broadest umbrella in the set: recent-version / remote_code breakage during model loading. It is the best representative of the cluster\u2019s dominant theme, though many items are still distinct bugs.", "canonical_pr_reason": null, - "best_issue_reason": "44509 is the strongest representative issue here: specific, actionable, and it captures the broader docs breakage around removed v5 pipeline tasks.", + "best_issue_reason": "45020 is the best global anchor because it captures the broadest and most central loading/regression pattern in the cluster and can contextually group several nearby reports, even if it does not subsume every issue here.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Both are compatibility/runtime errors, but they involve different failure modes: bitsandbytes init kwargs vs missing Float8 storage." + "reason": "Different bugs: bitsandbytes init-arg incompatibility vs missing Float8 storage registration." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Both are model-loading problems, but they target different models and different likely root causes." + "reason": "Both are model-loading failures, but for different models and different failure modes." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Both mention SigLIP2 behavior, but one is zero probabilities and the other is embedding quality; too broad to be the same bug." + "reason": "Same SigLIP2 family, but zero probabilities and degraded text embeddings are different symptoms without enough evidence of one root cause." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Different code paths: gradient checkpointing/LoRA versus int4 quantization memory/OOM." + "reason": "Both involve memory, but one is gradient checkpointing/LoRA ineffective and the other is int4 quantization OOM." }, { "left": "issue:43054", "right": "issue:43493", - "accept": false, - "reason": "Related SigLIP2 symptoms, but one is a quality complaint and the other is an HF-vs-JAX implementation discrepancy; not clearly the same defect." + "accept": true, + "reason": "Same SigLIP2 text-embedding regression/discrepancy; the HF-vs-JAX mismatch plausibly explains the degraded embeddings." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "Different model families and different failure signatures; both are load issues but not the same bug." + "reason": "Unrelated failures: GLM-4V loading vs Molmo missing all_tied_weights_keys." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "One is a generic noisy-loading complaint; the other is a specific tokenizer load failure for a particular model." + "reason": "Generic loading noise vs a specific model load failure; not the same underlying bug." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "Both touch training/memory, but one is about checkpointing effectiveness and the other about MoE training memory usage." + "reason": "Different Qwen3 MoE training issues with different causes and symptoms." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "Both involve remote_code, but 45020 is a broad regression report while 43643 is a specific AutoConfig field-loss bug." + "reason": "45020 is a broad umbrella; 43643 is a specific AutoConfig/trust_remote_code symptom, not clearly the same bug." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Different areas entirely: detokenization versus flash-attn import handling on Python 3.13." + "reason": "Different subsystems entirely: detokenization regression vs flash_attn import key error." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Different models and failure modes; no evidence they share the same code path." + "reason": "Different models and different load-time failures." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Qwen2.5-VL import error vs bitsandbytes _is_hf_initialized mismatch." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "Both are generic loading complaints, but they concern different models and likely unrelated issues." + "reason": "Both mention loading, but one is docs/UX noise while the other is a concrete model load failure." }, { "left": "issue:42491", "right": "issue:44368", "accept": false, - "reason": "Both mention Qwen3-family LoRA/training warnings, but one is a model-compatibility issue and the other is a tie_word_embeddings warning." + "reason": "Both are Qwen LoRA-related, but one is cross-version model usability and the other is a tie_word_embeddings warning." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Both are Gemma/Qwen loading failures, but they involve different models and different stack traces." + "reason": "Different DeepSpeed/Qwen problems: ZeRO-3 load failure vs FineGrainedFP8Config failure." }, { "left": "issue:43975", "right": "issue:44451", "accept": false, - "reason": "No concrete overlap beyond both being model-related regressions." + "reason": "Different model-specific tokenizer/detokenization issues." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "45020 is a broad remote_code regression; 45356 is a specific tokenizer codec/fix_mistral_regex issue." + "reason": "Tokenizer codec/warning regression is not the same as broad remote_code breakage." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "One is a tokenizer-mapping add-model helper failure; the other is a broader remote_code regression." + "reason": "Model-class mapping failure vs broad remote_code breakage; related area, not same bug." }, { "left": "issue:43054", "right": "issue:43278", "accept": false, - "reason": "Both discuss embeddings, but one is model quality and the other is dtype drift between train/eval." + "reason": "Both touch embeddings, but one is quality regression and the other is dtype drift between train/eval." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Different incompatibilities: missing Qwen2.5-VL import versus bitsandbytes init argument error." + "reason": "Same as above: import error vs bitsandbytes constructor mismatch." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer loading and flash-attn fallback checks are unrelated." + "reason": "Tokenizer loading regression vs flash-attn fallback gating; different code paths." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Both report the same v5 docs regression: pipeline task references that should no longer be present." + "reason": "Same docs regression: both still reference pipeline tasks removed in v5." }, { "left": "issue:43792", "right": "issue:44220", - "accept": false, - "reason": "The Whisper model load failure may be related to feature extraction, but the titles do not show the same concrete bug." + "accept": true, + "reason": "Both point to the Whisper audio feature-extraction/fbank path failing." }, { - "left": "issue:43404", - "right": "issue:43950", + "left": "issue:45081", + "right": "issue:45399", "accept": false, - "reason": "Different bugs: weight tying in Mistral3 versus non-persistent buffer corruption in from_pretrained." + "reason": "Tokenizer regex crash vs flash-attn fallback checks; unrelated failures." }, { - "left": "issue:45081", - "right": "issue:45399", + "left": "issue:43404", + "right": "issue:43950", "accept": false, - "reason": "Different tokenizer regex error versus flash-attn fallback gating." + "reason": "Different bugs: untied lm_head weights vs non-persistent buffer corruption in from_pretrained." }, { "left": "issue:43257", "right": "issue:43866", "accept": false, - "reason": "Different model/data issues: Qwen3 MoE weight conversion versus Ovis2 checkpoint corruption." + "reason": "Different models and unrelated loading/corruption reports." }, { "left": "issue:42915", "right": "issue:45020", "accept": false, - "reason": "A specific Qwen3 MoE FP8 failure is not the same as a broad remote_code regression." + "reason": "Qwen3 MoE FP8 training issue vs broad remote_code breakage; not the same underlying bug." }, { "left": "issue:43742", "right": "issue:45399", "accept": false, - "reason": "Different loading failures: MobileLLM key error versus flash-attn fallback checks." + "reason": "MobileLLM key error vs flash-attn fallback blocking; different subsystems." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "No concrete duplicate signal beyond both being model-loading regressions." + "reason": "Specific model key error is not the same as broad remote_code regression." }, { "left": "issue:41553", "right": "issue:44991", "accept": false, - "reason": "Different models and symptoms; one is an AutoTokenizer error, the other a tokenizer load regression for a specific model." + "reason": "Bad error message for one model vs tokenizer load failure for another; not duplicate." }, { "left": "issue:43940", "right": "issue:45313", "accept": false, - "reason": "Both concern Qwen3.5/Qwen3-Next ZeRO-3 weight loading, but the model targets and failure descriptions differ." + "reason": "Both are ZeRO-3 weight-loading failures, but different Qwen variants and likely different root causes." }, { "left": "issue:43872", "right": "issue:44291", "accept": false, - "reason": "Similar init/compatibility theme, but the specific error surfaces are different." + "reason": "Different bitsandbytes/init-empty-weights incompatibilities with different error signatures." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Both are model-loading issues, but they involve different models and no shared failure mechanism is evident." + "reason": "Different model-loading failures for different models." }, { "left": "issue:43257", "right": "issue:45305", "accept": false, - "reason": "Both are about Qwen3 training behavior, but one is weight conversion/load and the other is gradient averaging with GAS." + "reason": "Qwen3 MoE accelerate/deepspeed weight conversion vs DeepSpeed gradient averaging; related area, not same bug." }, { "left": "issue:41553", "right": "issue:44488", "accept": false, - "reason": "Different model-loading failures; no sign of a shared root cause." + "reason": "Different models and different failure types." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Completely different problems: Trainer compute_loss keyword mismatch versus tensor mask shape mismatch." + "reason": "Trainer compute_loss API mismatch vs tensor mask shape mismatch; unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The single accepted pair is well grounded: both issues describe the same v5 docs regression around removed pipeline-task references, and they look mergeable into one docs-fix PR. The other verdicts are conservative and do not overstate duplicate evidence." + "feedback": "The summary is grounded in the packet and stays conservative about the cluster being a mix of unrelated loading/regression issues. The only accepted soft edges are well-supported by the titles (docs pipeline removals, SigLIP2 discrepancy, Whisper fbank path), and the rejected pairs are generally framed cautiously rather than overstated. 45020 is a reasonable broad anchor, with appropriate caveats that it does not subsume the whole cluster." }, "evaluator_used": true, "retried": false, @@ -1301,227 +1361,227 @@ "issue:43596|issue:44661", "issue:43644|issue:44534", "issue:43317|issue:44756", - "issue:38175|issue:39692", "issue:45313|issue:45357", + "issue:38175|issue:39692", "issue:42491|issue:43720", "issue:44960|issue:44977", "issue:44451|issue:44488", "issue:43708|issue:44075" ], - "prepared_review_unit_hash": "52d3cb6a2107ec41cee5d1b2aa982ca929505883d3d61a539ae14b838086f9f2", + "prepared_review_unit_hash": "31e1998fd2fad8db2d1200aec73d9372eb673c7520456010259c4dbe642fad94", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36649, - "estimated_input_tokens": 9163, - "estimated_eval_tokens": 18582 + "serialized_chars": 36648, + "estimated_input_tokens": 9162, + "estimated_eval_tokens": 18580 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous, but a few duplicate pockets are real: the non-persistent-buffer regressions, the custom-model init regression, and some Qwen/SigLIP2 output/video reports. The soft edges should stay conservative where only a shared model family or broad subsystem is evident.", - "confidence": 0.57, - "canonical_issue_reason": "Issue 43950 is the cleanest, most precisely scoped regression report in the set and has an exact near-duplicate in 44534, making it a strong canonical representative for the non-persistent-buffer bug family.", + "summary": "Mostly heterogeneous issue set. I\u2019d keep only two soft duplicate pairs as accepted: the mask-shape mismatch reports and the non-persistent-buffer corruption reports. The other pairs are related by model family or broad subsystem in some cases, but the evidence here is not strong enough to treat them as the same underlying bug.", + "confidence": 0.77, + "canonical_issue_reason": "Issue 43644 is the clearest canonical report for the non-persistent-buffer regression: it describes the Transformers v5 junk-fill behavior directly, and 44534 appears to be the same bug with a near-identical title.", "canonical_pr_reason": null, - "best_issue_reason": "43950 is the best overall issue to anchor this cluster because it is concrete, regression-oriented, and has the strongest duplicate evidence among the issues shown.", + "best_issue_reason": "Issue 43644 is the strongest representative issue in this set because it is explicit, actionable, and has a close duplicate in 44534; it best anchors the non-persistent-buffer regression cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41093", "right": "issue:44805", "accept": true, - "reason": "Both report the same mask-vs-tensor shape mismatch IndexError with nearly identical wording, so they look like the same underlying bug." + "reason": "Same mask/index shape-mismatch failure; 44805 is the broader wording of the same IndexError." }, { "left": "issue:44479", "right": "issue:44560", - "accept": true, - "reason": "Both are Qwen3.5/Qwen3-VL video-input regressions in the same version range, and the failure descriptions are close enough to plausibly be the same root issue." + "accept": false, + "reason": "Both involve Qwen video paths, but the titles point to different regressions and not one concrete shared bug." }, { "left": "issue:43872", "right": "issue:45538", "accept": false, - "reason": "These are unrelated failures: bitsandbytes initialization incompatibility versus CLIPTokenizer max-length handling." + "reason": "Different subsystems and failures: bitsandbytes storage/init incompatibility vs tokenizer max-length behavior." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "Different model families and different failure modes; the packet does not support a shared underlying bug." + "reason": "Unrelated problems: PagedAttentionCache/Qwen3.5 crash vs NemotronH checkpoint loading." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "One is GGUF/gpt-oss architecture support, the other is a torch.autocast dtype mismatch on Phi-tiny-MoE; too different to treat as duplicates." + "reason": "Different bugs: GGUF/gpt-oss support vs Phi-tiny-MoE dtype mismatch under autocast." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "DeepSpeed training weight-shape error and image-processor segmentation-map handling are distinct code paths and problems." + "reason": "No shared code path: DeepSpeed weight-shape error vs void segmentation-map processing." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "These only share the broad Qwen/GGUF area; one is architecture support, the other is a sliding_window bug." + "reason": "Only a loose Qwen3-MoE association; one is GGUF support, the other is a sliding-window bug." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Both touch attention-related paths, but the concrete failures and model contexts differ and there is no evidence of the same regression." + "reason": "Different model families and mechanisms: Bamba torch.compile/SDPA vs Qwen3.5 flash-attention position_ids handling." }, { "left": "issue:43645", "right": "issue:43646", - "accept": true, - "reason": "Both describe the Transformers 5.0 custom-model initialization regression, with 43645 just being the notebook variant of 43646." + "accept": false, + "reason": "Related Transformers 5.0 custom-model breakage, but the reported failure modes are not specific enough to merge as one bug." }, { "left": "issue:43950", "right": "issue:45325", "accept": false, - "reason": "Non-persistent-buffer corruption and Qwen2.5-VL rope/position-id scaling are different bugs." + "reason": "Different regressions: non-persistent-buffer corruption vs Qwen2.5-VL rope/index scaling." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "GGUF support and a Float8 storage lookup error are unrelated." + "reason": "Broad model-loading theme only; gpt-oss GGUF support is unrelated to missing Float8 storage." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Both involve Qwen/video paths, but one is a StopIteration in video handling and the other is SFT embedding unpacking; not enough evidence of the same bug." + "reason": "Different failures in different paths: video StopIteration vs embedding unpacking during TRL SFT training." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Missing import symbol and meta-device loading breakage are different failures." + "reason": "Both are load/import regressions, but they affect different classes of models and code paths." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "Qwen3MoE FP8 config failure and meta-device loading regression are distinct issues." + "reason": "Qwen3Moe FP8 failure and meta-device loading issues are distinct defects." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "3d_parallel.py execution failure is unrelated to GGUF gpt-oss support." + "reason": "3d_parallel.py runtime issue is unrelated to GGUF gpt-oss support." }, { "left": "issue:38175", "right": "issue:43994", - "accept": true, - "reason": "Both are about bad output from the same SigLIP2 model family, and the titles point to a closely related model/processor regression." + "accept": false, + "reason": "Same model family and symptom class, but the evidence only shows similar bad outputs, not the same concrete bug." }, { "left": "issue:43646", "right": "issue:43950", "accept": false, - "reason": "Custom model initialization and non-persistent-buffer corruption are different regressions." + "reason": "Custom model initialization breakage and non-persistent-buffer corruption are different regressions." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "The video-input StopIteration report and the mask-shape IndexError do not look like the same underlying code-path bug." + "reason": "Video StopIteration and mask/tensor shape mismatch are unrelated failures." }, { "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "GenerationMixin import failure and Qwen3MoE FP8 failure are unrelated." + "reason": "ImportError for GenerationMixin is unrelated to Qwen3Moe FP8 config failure." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "DeepSpeed/BertModel zero3 initialization and tokenizer-mapping-name handling are separate problems." + "reason": "Different code paths: DeepSpeed ZeRO-3 empty-index failure vs tokenizer-mapping-name handling." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Both involve config loading, but the packet does not show they are the same regression." + "reason": "Both are config-loading regressions, but they target different models and validation failures." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Zero3/BertModel index error and add-new-model-like tokenizer mapping failure are different issues." + "reason": "DeepSpeed ZeRO-3 BertModel failure is unrelated to add-new-model-like/tokenizer mapping." }, { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "These are the same non-persistent-buffer junk-fill regression, phrased as a direct duplicate." + "reason": "Near-identical reports of the same Transformers v5 non-persistent-buffer junk-fill regression." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "GPU+CPU offload loading failure and Strix Halo mmap/OOM avoidance are different loading/memory issues." + "reason": "Different loading/memory issues: dequantized offload loading vs Strix Halo mmap/OOM." }, { - "left": "issue:38175", - "right": "issue:39692", + "left": "issue:45313", + "right": "issue:45357", "accept": false, - "reason": "They only share the SigLIP2 family; one is about zero probabilities while the other is a docs-example/model-processor/quantization bundle of errors, so the evidence is too weak for duplicate status." + "reason": "Separate Qwen3.5 regressions: ZeRO-3 weight loading vs incorrect save_pretrained visual-encoder keys." }, { - "left": "issue:45313", - "right": "issue:45357", + "left": "issue:38175", + "right": "issue:39692", "accept": false, - "reason": "DeepSpeed ZeRO-3 weight loading and save_pretrained key serialization are different Qwen3.5 regressions." + "reason": "Both involve SigLIP2, but one is output quality and the other is a docs/example mismatch plus quantization failure." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "LoRA compatibility on qwen3_moe and packed-weight unpacking during accelerate loading are distinct loading paths and bugs." + "reason": "Different model-loading regressions: Qwen3-MoE LoRA compatibility vs BitNet packed-weight unpacking." }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "GLM5 and Qwen3.5 flash-attention generation failures are unrelated." + "reason": "Unrelated models and failures: GLM5 vs Qwen3.5 flash-attention generation bug." }, { "left": "issue:44451", "right": "issue:44488", "accept": false, - "reason": "Both are BERT-loading complaints, but the packet does not show they stem from the same regression or model-loading path." + "reason": "Both are model-load failures, but they point to different checkpoints and no shared defect is established." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Checkpoint-resume max_steps calculation and SGD argument handling are unrelated trainer/optimizer issues." + "reason": "Trainer resume/max_steps accounting bug is unrelated to SGD argument handling." } ] }, "evaluator_result": { "accept": false, - "feedback": "The summary is mostly grounded, but a few soft-edge accepts are too strong for the evidence in the packet. In particular, issue:38175 vs issue:39692 only share the SigLIP2 family; the titles point to different failure modes (zero probabilities vs docs-example/model-processor/quantization problems), so calling them likely duplicates overstates the record. Likewise issue:44451 vs issue:44488 are both BERT-loading complaints, but the packet doesn\u2019t show they\u2019re the same regression. Because the analyst isn\u2019t consistently conservative on these edges, I can\u2019t accept the output." + "feedback": "The writeup is mostly grounded, but it overstates duplicate certainty in a couple of soft-edge calls. In particular, issue:44479 vs issue:44560 and issue:38175 vs issue:43994 are only loosely supported by the titles alone; they look related by model family / symptom class, but not clearly the same bug. The summary also phrases these as \"clear duplicate pairs,\" which is stronger than the packet evidence justifies. The safer accepted pairs are issue:41093 vs issue:44805 and issue:43644 vs issue:44534; the rest should stay conservative." }, "evaluator_used": true, "retried": true, @@ -1595,9 +1655,9 @@ "issue:43854|issue:43931", "issue:43856|issue:43873", "issue:42491|issue:45276", - "issue:39692|issue:43994", "issue:41720|issue:42915", "issue:43257|issue:43854", + "issue:39692|issue:43994", "issue:43856|issue:44387", "issue:44488|issue:44991", "issue:45362|issue:45406", @@ -1605,9 +1665,9 @@ "issue:42915|issue:44756", "issue:42491|issue:43931", "issue:44661|issue:44683", - "issue:43404|issue:43454", - "issue:42491|issue:43856", "issue:45276|issue:45335", + "issue:42491|issue:43856", + "issue:43404|issue:43454", "issue:36010|issue:39404", "issue:44164|issue:44291", "issue:41720|issue:44155", @@ -1634,287 +1694,290 @@ "issue:43425|issue:43828", "issue:43828|issue:43957" ], - "prepared_review_unit_hash": "d23050f29ab58e6d69837298c099b7b83bd76c0e853a7655a2012de1b31f0d97", + "prepared_review_unit_hash": "1a6e86bc21033eaca11499ada59c91691e59c965eadb7f5b81a8eb94ddf772af", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 41, - "serialized_chars": 39804, - "estimated_input_tokens": 9951, - "estimated_eval_tokens": 20158 + "serialized_chars": 39812, + "estimated_input_tokens": 9953, + "estimated_eval_tokens": 20162 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a collection of distinct Transformers regressions across different model families and subsystems, not one duplicate set. A few subgroups are thematically close (e.g. custom model initialization, resize_token_embeddings, Qwen3/MoE loading), but the soft pairs still look like separate concrete bugs rather than the same underlying change.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 43646 is the broadest representative of the initialization/regression subgroup and is more general than nearby variants like 43645, 44164, and 44291, but the overall cluster is too heterogeneous to treat as a single duplicate chain.", + "summary": "The batch contains multiple loose groups, with at least two coherent duplicate families: a resize_token_embeddings propagation family (45292/45276/45335) and a separate lm_head weight-tying family (43404/43454). 45292 is the best representative for the resize_token_embeddings family, but the cluster is not single-family.", + "confidence": 0.79, + "canonical_issue_reason": "issue:45292 is the best representative of the resize_token_embeddings family because it describes the generic propagation failure most clearly, with 45276 and 45335 as close model-specific variants.", "canonical_pr_reason": null, - "best_issue_reason": "43646 is the best single issue to represent the cluster if one must be chosen: it has the broadest scope and best matches the recurring Transformers 5.x model-initialization/load regression theme. It is still only a partial fit because many other items are separate bugs.", + "best_issue_reason": "issue:45292 is the strongest issue candidate for one coherent duplicate family, but not the only one; the batch also contains a plausible separate lm_head weight-tying family around 43404/43454.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Both are model-loading failures, but for different models and different symptoms (unit-test load issue vs shape mismatch)." + "reason": "Different model-loading failures; same broad area, not the same bug." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Same quantization/memory neighborhood, but one is Qwen3 MoE training memory inefficiency while the other is offloading behavior with quantization." + "reason": "Both involve memory/quantization, but the symptoms and code paths differ." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "Both mention Qwen3/MoE-style model behavior, but one is a LoRA compatibility break and the other is a token embedding resize bug." - }, - { - "left": "issue:39692", - "right": "issue:43994", - "accept": false, - "reason": "Both concern SigLIP2/Siglip2 usage, but one is a broken documentation example and the other is an inference-quality issue with AutoModel/pipeline." + "reason": "LoRA checkpoint compatibility vs resize_token_embeddings propagation are unrelated." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Different failure modes: CUDA assert on auto device mapping versus FineGrainedFP8Config load failure." + "reason": "Different Qwen3 failure modes: cuda assert vs FP8 config loading." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Qwen3 MOE weight conversion under accelerate+deepspeed is a different path from loading GLM-4.7-Flash in tests." + "reason": "Separate model-loading bugs for different models and mechanisms." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Same model family, but one is a docs/example issue and the other is an inference correctness problem." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both involve memory/quantization, but one is training-memory inefficiency and the other is int4 CUDA reserved-memory/OOM regression." + "reason": "Both mention memory pressure, but they are distinct regressions." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Different models and different load paths; one is a BERT checkpoint load failure, the other is tokenizer loading for a specific RoBERTa model." + "reason": "Both are tokenizer-load failures, but for different models with no clear shared concrete bug." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Both are chat/processor related, but one is a Qwen3.5 crash and the other is a Gemma4Processor missing `_tokenizer` in transformers serve." + "reason": "Completely different components: chat crash vs processor attribute error." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Both are Transformers 5.0 regressions, but one is junk non-persistent buffers and the other is custom model initialization in Jupyter." + "reason": "Different Transformers 5.0 regressions with different failure surfaces." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "Both touch memory behavior, but one is FP8 config load failure and the other is a platform-specific mmap/OOM issue." + "reason": "FP8 loading failure vs mmap/OOM workaround are unrelated." }, { "left": "issue:42491", "right": "issue:43931", "accept": false, - "reason": "Different regressions: LoRA model portability across versions versus Qwen3-VL weight-shape mismatch." + "reason": "Different Qwen3 checkpoint/model-loading problems." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "Different subsystems: tokenizer mapping registration versus compiled flex_attention on newer torch." + "reason": "Tokenizer mapping bug vs torch flex_attention compile failure are unrelated." }, { - "left": "issue:43404", - "right": "issue:43454", - "accept": false, - "reason": "Both are weight-tying bugs, but for different model classes (Mistral3 vs AyaVision) and different reported effects." + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Same resize_token_embeddings regression: resized embeddings are not propagated to model-specific output/decoder embeddings." }, { "left": "issue:42491", "right": "issue:43856", "accept": false, - "reason": "LoRA version-compatibility and Qwen3 MoE training memory are unrelated bugs." + "reason": "Different Qwen3 MoE checkpoint compatibility and memory-usage problems." }, { - "left": "issue:45276", - "right": "issue:45335", - "accept": false, - "reason": "These are very close, but still model-specific reports of the same resize_token_embeddings family rather than a clearly identical bug with the same code path." + "left": "issue:43404", + "right": "issue:43454", + "accept": true, + "reason": "Same lm_head weight-tying defect showing up in two model classes; one fix could plausibly cover both." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "ImportError for GenerationMixin is unrelated to Whisper return_language pipeline behavior." + "reason": "ImportError for GenerationMixin and Whisper pipeline return_language are unrelated." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "Both are save/load related, but one is extra_state handling and the other is init_empty_weights getting an unexpected argument." + "reason": "Different serialization/init regressions: extra_state handling vs init_empty_weights argument mismatch." }, { "left": "issue:41720", "right": "issue:44155", "accept": false, - "reason": "CUDA assert during Qwen3 auto device mapping is unrelated to AudioFlamingo3 batched inference token leakage." + "reason": "Different models and failures; no shared code-path evidence." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "Different model-loading failures with different root causes and checkpoints." + "reason": "Both are loading/tokenizer-related, but the reported model-specific failures differ." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Quantization/offloading expectations differ from the specific int4 reserved-memory/OOM regression." + "reason": "Quantization/offloading memory issues, but not the same concrete defect." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "SigLIP2 documentation errors are unrelated to the GLM5 issue." + "reason": "SigLIP2 example bug and GLM5 are unrelated." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Different memory bugs: int4 quantization OOM versus disabling mmap on Strix Halo." + "reason": "Both are OOM-related, but the triggers and fixes are unrelated." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch 2.10 incompatibility is a version-compatibility issue; the other is meta-device loading breaking some models." + "reason": "Torch version compatibility and meta-device loading are distinct problems." }, { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Both are model-loading/quantization adjacent, but one is Qwen-3-8B-NVFP4 runtime failure and the other is MXFP4 falling back to bf16." + "reason": "Different quantization formats and model families." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Gradient checkpointing with PEFT LoRA and gradient checkpointing in eval mode are related only at a very high level, not the same bug." + "reason": "Different gradient-checkpointing issues with different conditions." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "Both are Qwen-ish loading regressions, but one is GLM-4.7-Flash load issues while the other is a Qwen3.5 save_pretrained checkpoint regression." + "reason": "Different models and different failure modes: loading vs save_pretrained regression." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "Unexpected `_is_hf_initialized` during meta init is a model construction bug, not a platform-specific mmap OOM issue." + "reason": "Unexpected init argument vs Strix Halo mmap/OOM are unrelated." }, { "left": "issue:45276", "right": "issue:45292", - "accept": false, - "reason": "Very close topic, but 45276 is a model-specific report while 45292 is the broader generic resize_token_embeddings/output_embeddings issue." + "accept": true, + "reason": "Same resize_token_embeddings family; 45292 is the broad generic report and 45276 is a model-specific manifestation." }, { "left": "issue:44743", "right": "issue:44960", "accept": false, - "reason": "Qwen3.5 recurrent-state reset in modular code is unrelated to the GLM5 loading thread." + "reason": "Recurrent-state cache bug and GLM5 are unrelated." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Eval-mode checkpointing restriction is unrelated to the RLHF NaN/position_ids SDPA fallback bug." + "reason": "Gradient-checkpointing eval-mode restriction vs RLHF NaN explosion are different bugs." }, { "left": "issue:43957", "right": "issue:44756", "accept": false, - "reason": "Meta-device loading breakage is a different mechanism from disabling mmap to avoid OOM." + "reason": "Meta-device loading failure vs mmap/OOM workaround are unrelated." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "Extra_state save/load handling is unrelated to mmap-based memory pressure." + "reason": "Serialization extra_state bug vs Strix Halo memory issue are unrelated." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom model initialization in Transformers 5.0 is unrelated to resume_from_checkpoint max_steps calculation." + "reason": "Custom model initialization regression and resume_from_checkpoint max_steps bug are unrelated." }, { "left": "issue:45292", "right": "issue:45335", - "accept": false, - "reason": "These are same topic but still separate model-specific reports; insufficient evidence they are the exact same underlying defect." + "accept": true, + "reason": "Same resize_token_embeddings propagation bug affecting output/decoder embeddings." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Video-input regression in several Qwen variants is unrelated to strict config blocking granite_speech." + "reason": "Video-input regression and strict config loading are unrelated." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "Different checkpoint/model load failures for different model names." + "reason": "Tokenizer vocab mismatch and model load failure are not clearly the same defect." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Jupyter custom model init regression is not the same bug as save/from_pretrained extra_state handling." + "reason": "Notebook/custom-model init issue and extra_state save/load issue are separate regressions." }, { "left": "issue:43257", "right": "issue:44863", "accept": false, - "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed is unrelated to NemotronH checkpoint loading." + "reason": "Qwen3 MoE deepspeed loading and NemotronH checkpoint loading are different model-specific bugs." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch 2.10 incompatibility is a general version issue; the other is a dtype mismatch under autocast for Phi-tiny-MoE." + "reason": "Torch compatibility issue and autocast dtype mismatch are unrelated." }, { "left": "issue:43828", "right": "issue:43957", "accept": false, - "reason": "Autocast dtype mismatch and meta-device loading breakage are different code paths and symptoms." + "reason": "Autocast dtype mismatch and meta-device loading failure are different bugs." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, - "accepted_nontrivial_soft_edge": false, + "evaluator_result": { + "accept": false, + "feedback": "Reject: the packet supports at least two coherent duplicate families, not just one. The resize_token_embeddings set is grounded, but the best_issue_reason overstates uniqueness by calling it the only strongly coherent duplicate set. Also, the lm_head weight-tying pair (43404/43454) is itself a plausible duplicate family, so the summary should be more conservative about cluster structure." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -1984,13 +2047,13 @@ "issue:44589|issue:44756", "issue:43906|issue:44530", "issue:43931|issue:44863", - "issue:43611|issue:43645", "issue:39404|issue:44912", + "issue:43611|issue:43645", "issue:44530|issue:45084", "issue:43425|issue:43716", "issue:44265|issue:44898", - "issue:43299|issue:43931", "issue:43901|issue:44509", + "issue:43299|issue:43931", "issue:44479|issue:44805", "issue:43994|issue:44079", "issue:44671|issue:44991", @@ -2017,232 +2080,232 @@ "issue:34567|issue:40444", "issue:43716|issue:45084" ], - "prepared_review_unit_hash": "35bb086639708de5695cf19f327277da14d9fd83b77df03dabeab28ee77736e6", + "prepared_review_unit_hash": "d315a8e1dee27dd5881a7440280ae534ce3c476401cce9376c2704d404937530", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 38176, - "estimated_input_tokens": 9544, - "estimated_eval_tokens": 19344 + "serialized_chars": 38181, + "estimated_input_tokens": 9546, + "estimated_eval_tokens": 19348 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose grab-bag of mostly unrelated issues, with many v5 regression reports but different code paths (model loading, tokenizers, audio, vision, MoE, docs, and training utilities). I do not see any soft-edge pair that is clearly the same underlying bug/change.", - "confidence": 0.68, - "canonical_issue_reason": "issue:43611 is the broadest and most representative v5 regression report here, centered on model loading breaks and base_model_prefix behavior.", + "summary": "This cluster is mostly a heterogeneous grab-bag of Transformers v5 regression reports, with many unrelated model/tokenizer/training issues. I do not see any soft-edge pair that looks like a true duplicate, so all soft edges are rejected.", + "confidence": 0.72, + "canonical_issue_reason": "issue:43611 is the broadest, most general representative here: a Transformers v5 model-loading regression. It best fits the recurring pattern across the cluster better than the narrower model- or tokenizer-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the best single representative issue because it captures the main release-regression theme while staying concrete.", + "best_issue_reason": "issue:43611 is the best umbrella issue for this cluster because it describes the most general load-path regression and can absorb the broad duplicate chatter around v5 loading failures.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Different failures: float8 storage lookup vs Apple Silicon mmap/OOM behavior." + "reason": "Different failure classes: storage lookup/type error vs mmap-related OOM avoidance. No shared code path." }, { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "Unrelated code paths: isolated repro of an older issue vs PagedAttentionCache linear_attention crash." + "reason": "An isolated reproduction of a prior issue is not the same as a PagedAttentionCache linear_attention crash; different models and code paths." }, { "left": "issue:43931", "right": "issue:44863", "accept": false, - "reason": "Both are loading issues, but for different models and different failure modes (shape mismatch vs checkpoint load support)." + "reason": "Both are model-loading complaints, but they target different model families and failure modes (shape mismatch vs checkpoint loading support)." }, { - "left": "issue:43611", - "right": "issue:43645", + "left": "issue:39404", + "right": "issue:44912", "accept": false, - "reason": "Same release family, but different concrete bugs: base_model_prefix loading vs custom model init in Jupyter." + "reason": "Whisper pipeline language-return regression is unrelated to MXFP4 quantization fallback/loading behavior." }, { - "left": "issue:39404", - "right": "issue:44912", + "left": "issue:43611", + "right": "issue:43645", "accept": false, - "reason": "Whisper pipeline return_language regression is unrelated to GPT-OSS MXFP4 quantization fallback." + "reason": "Both involve v5 loading regressions, but one is base_model_prefix/model loading and the other is custom model initialization in notebooks; not the same bug." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "Different subsystems: Qwen3.5 cache group-type error vs template compilation error." + "reason": "Qwen3.5 cache crash and template-node compilation error are unrelated issues." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Torch version incompatibility is not the same as a Mistral image-preprocessor dtype mismatch." + "reason": "Torch version incompatibility and an image preprocessor dtype mismatch are separate problems." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export/torch_compilable_check bug is unrelated to Perceiver interpolation at non-default resolution." + "reason": "torch.export with torch_compilable_check is a different failure path from Perceiver interpolation handling." }, { - "left": "issue:43299", - "right": "issue:43931", + "left": "issue:43901", + "right": "issue:44509", "accept": false, - "reason": "Different Qwen3-VL-related failures with different root causes; not the same loading bug." + "reason": "Both are docs issues, but they cover different documentation gaps and different pipeline/task content." }, { - "left": "issue:43901", - "right": "issue:44509", + "left": "issue:43299", + "right": "issue:43931", "accept": false, - "reason": "Both are docs-related, but they document different pipeline behaviors and different removals." + "reason": "Same broad area (Qwen/VL loading) but different concrete bugs: MoE loading breakage vs checkpoint shape mismatch." }, { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Video-input regression for Qwen VL models is unrelated to a mask shape mismatch error." + "reason": "Video-input regression and mask-shape IndexError are not the same underlying bug." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "Incorrect Siglip2 outputs and ModelOutput key assignment are different bugs." + "reason": "One is a bad-results complaint for a specific model; the other is a ModelOutput key assignment bug." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "CamemBERT MLM prediction issues and tokenizer loading failure are separate problems." + "reason": "CamemBERT prediction correctness and tokenizer loading for est-roberta are unrelated." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "Qwen3 MoE memory usage and Strix Halo mmap/OOM are unrelated performance/pathology reports." + "reason": "MoE training memory inefficiency is unrelated to disabling mmap on Strix Halo to avoid OOM." }, { "left": "issue:39404", "right": "issue:44671", "accept": false, - "reason": "Whisper pipeline regression and CamemBERT predictions are clearly different models and code paths." + "reason": "Whisper pipeline language output and CamemBERT masked-LM predictions are different model-specific regressions." }, { "left": "issue:40990", "right": "issue:42947", "accept": false, - "reason": "Perplexity anomaly and LoRA gradient-checkpointing behavior do not describe the same bug." + "reason": "Perplexity anomaly and gradient-checkpointing ineffectiveness are distinct training/optimization issues." }, { "left": "issue:44589", "right": "issue:45084", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to template-node compilation errors." + "reason": "Float8 storage lookup error and template-node compilation failure are unrelated." }, { "left": "issue:43596", "right": "issue:44683", "accept": false, - "reason": "DeepSpeed ZeRO-3 init index error and compiled flex_attention failure are separate issues." + "reason": "DeepSpeed ZeRO-3 initialization index error and compiled flex_attention on torch>=2.9 are different code paths." }, { "left": "issue:43425", "right": "issue:44918", "accept": false, - "reason": "Torch 2.10 incompatibility is unrelated to Qwen3.5 embedding unpacking with TRL SFT." + "reason": "Torch 2.10 incompatibility and TRL input-embedding unpacking failure are unrelated regressions." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "Tokenizer method-name conflict is not the same as gguf_file breaking auto loading." + "reason": "T5 tokenizer special-token conflict and gguf_file/from_pretrained breakage are not the same bug." }, { "left": "issue:32090", "right": "issue:39404", "accept": false, - "reason": "Trainer GPU broadcast NoneType error is unrelated to Whisper pipeline return_language behavior." + "reason": "Trainer GPU broadcast NoneType error and Whisper return_language pipeline regression are unrelated." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError and _is_hf_initialized flag break are different failures." + "reason": "Apple Silicon warmup TypeError and _is_hf_initialized flag breakage are different issues." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Tokenizer loading for est-roberta and Mistral regex patch crashes are different tokenizer bugs." + "reason": "Both are tokenizer-loading problems, but one is a specific tokenizer load failure while the other is a Mistral regex patch crash with a different trigger." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Perplexity regression and runtime special-token post-processor updates are unrelated." }, { "left": "issue:43957", "right": "issue:44164", "accept": false, - "reason": "Meta-device loading regressions and extra_state save/load handling are distinct serialization paths." + "reason": "Meta-device loading breakage and extra_state save/from_pretrained handling are separate model lifecycle bugs." + }, + { + "left": "issue:44164", + "right": "issue:44479", + "accept": false, + "reason": "Extra_state serialization and video-input regression have no common concrete failure path." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Loading cjvt/sleng-bert and patch_mistral_regex backend_tokenizer crash are unrelated." + "reason": "Tokenizer load failure for a specific model and Mistral regex patch crash are different issues." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Missing all_tied_weights_keys on MolmoForCausalLM is unrelated to sleng-bert tokenizer loading." + "reason": "Missing all_tied_weights_keys on Molmo and tokenizer-loading failure for sleng-bert are unrelated." }, { "left": "issue:43425", "right": "issue:45161", "accept": false, - "reason": "Torch version compatibility and GPT-OSS tensor-parallel loading are different concerns." + "reason": "Torch version incompatibility and GPT-OSS MoE tensor-parallel behavior are not the same bug." }, { "left": "issue:43611", "right": "issue:43644", "accept": false, - "reason": "Both are v5 regressions, but one is model-loading/base_model_prefix and the other is non-persistent buffer initialization." + "reason": "Both are v5 regressions, but one is model-loading/base_model_prefix and the other is non-persistent buffer initialization; different concrete problems." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "Both touch audio features, but the reported failures are not clearly the same concrete bug." + "reason": "Audio feature-extractor center-arg crash and _torch_extract_fbank_features() issue are related area-wise, but not clearly the same underlying defect." }, { "left": "issue:42947", "right": "issue:44756", "accept": false, - "reason": "LoRA checkpointing inefficiency and Strix Halo mmap/OOM are unrelated resource issues." + "reason": "LoRA gradient-checkpointing behavior and Strix Halo mmap OOM are unrelated." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "Different model/tokenizer loading failures with no shared code path." - }, - { - "left": "issue:34567", - "right": "issue:40444", - "accept": false, - "reason": "TrainerState token counting and Qwen2.5-VL iterable dataset multi-image finetuning are unrelated." - }, - { - "left": "issue:43716", - "right": "issue:45084", - "accept": false, - "reason": "Mistral-3 image dtype mismatch and template compilation failure are unrelated." + "reason": "Loading sleng-bert and loading physical-intelligence/fast on newer transformers are different tokenizer/model-loading failures." } ] }, @@ -2335,9 +2398,9 @@ "issue:45084|issue:45507", "issue:43994|issue:44610", "issue:43377|issue:43819", - "issue:43322|issue:43540", "issue:43645|issue:44534", "issue:45216|issue:45357", + "issue:43322|issue:43540", "issue:44403|issue:45081", "issue:44451|issue:45081", "issue:44265|issue:44610", @@ -2351,247 +2414,247 @@ "issue:43296|issue:43531", "issue:40990|issue:44811" ], - "prepared_review_unit_hash": "42e41e70fd77d3595fed7209c258f9eb0cdd11a305e5a5477412c4eb16f1a532", + "prepared_review_unit_hash": "15f4b920270d163dc4a5f4a0c6aac4791d3343a0d6bd795100e2495900d44ee6", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37882, - "estimated_input_tokens": 9471, - "estimated_eval_tokens": 19198 + "serialized_chars": 37892, + "estimated_input_tokens": 9473, + "estimated_eval_tokens": 19202 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly heterogeneous, but there are two clear duplicate-style pairs: the Transformer v5 non-persistent-buffer regression and the Qwen3.5 save_pretrained regression. Most other soft matches are only broad subsystem or model-family similarities, not the same bug.", - "confidence": 0.71, - "canonical_issue_reason": "Issue 43644 is the cleanest canonical representative of the exact duplicate pair about v5 filling non-persistent buffers with junk; it is the more specific original wording.", + "summary": "This cluster is too heterogeneous to treat as one duplicate set. The GGUF-loading and Qwen3.5 save_pretrained pairs look like real duplicates, but the Qwen video/rope pair is only loosely related and should not be accepted; overall the cluster should be rejected.", + "confidence": 0.94, + "canonical_issue_reason": "45216 is the clearest representative of the strongest subcluster: a precise Qwen3.5 save_pretrained regression with a direct follow-up in 45357.", "canonical_pr_reason": null, - "best_issue_reason": "43644 is the best issue to anchor on because it states the bug most directly and matches its near-duplicate almost verbatim.", + "best_issue_reason": "45216 is the best single issue in the packet because it is concrete, well-scoped, and has the strongest duplicate linkage.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43644", - "right": "issue:44534", - "accept": true, - "reason": "Same concrete regression; the titles describe the same v5 non-persistent-buffer junk initialization bug." + "right": "issue:43646", + "accept": false, + "reason": "Both are Transformers v5 regressions, but the bugs are different: non-persistent buffers vs custom model initialization." }, { "left": "issue:43883", "right": "issue:44991", "accept": false, - "reason": "Both are loading failures, but they involve different models and different failure modes." + "reason": "Both are loading failures, but they affect different models and different failure modes." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "Different bugs: missing fields from AutoConfig vs an invalid group type crash in PagedAttentionCache." + "reason": "AutoConfig missing fields and a PagedAttentionCache crash are unrelated code paths." }, { "left": "issue:43452", "right": "issue:43482", - "accept": false, - "reason": "Both mention GGUF loading, but one is a generic gguf_file/from_pretrained breakage and the other is a specific Qwen2.5-GGUF v5 failure." + "accept": true, + "reason": "Both are GGUF-loading regressions; 43482 looks like a model-specific instance of the broader `gguf_file` breakage." }, { "left": "issue:36032", "right": "issue:42915", "accept": false, - "reason": "Unrelated issues: T5 tokenizer special-token attribute conflict vs Qwen3-MoE FP8 config failure." + "reason": "Tokenizer method conflict and Qwen3Moe FP8 failure are unrelated." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "Different code paths and symptoms; device_map/offload loading is not the same as GGUF loading under v5." + "reason": "Device-map/offload dequantization failure is separate from GGUF loading." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Audio chat-template sampling-rate default bug vs MIMI encoder batching/padding-mask mismatch." + "reason": "Audio chat-template sampling-rate defaults and MIMI padding-mask batching are different bugs." }, { "left": "issue:44451", "right": "issue:44987", "accept": false, - "reason": "Both are model-loading regressions, but they are different models and likely different causes." + "reason": "Similar loading complaints, but no evidence of the same root cause." }, { "left": "issue:43541", "right": "issue:45084", "accept": false, - "reason": "Torch Dynamo grouped_mm tracing error is unrelated to the non-template-node compilation TypeError." + "reason": "Torch dynamo grouped_mm tracing error and template compilation TypeError are different failures." }, { "left": "issue:39404", "right": "issue:43825", "accept": false, - "reason": "Whisper pipeline return_language regression is distinct from a pipeline error-message wording issue." + "reason": "Whisper return_language behavior and a pipeline error-message issue are distinct." }, { "left": "issue:42915", "right": "issue:44530", "accept": false, - "reason": "Different failures in Qwen models: FP8 config loading vs PagedAttentionCache group-type crash." + "reason": "Qwen3Moe FP8 config failure and Qwen3.5 cache/group-type crash are different problems." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Different Qwen3.5 bugs: hidden-states output regression vs Qwen2.5-VL rope/position-id scaling." + "reason": "Different Qwen VL bugs: output_hidden_states behavior vs rope-index scaling for still-image temporal positions." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Related to tied weights, but one is a missing tie in Mistral3ForConditionalGeneration and the other is device-dependent serialization behavior." + "reason": "Both involve tied weights, but one is a missing tie and the other is a save/serialization inconsistency." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "T5 tokenizer special-token conflict is unrelated to custom model initialization in notebooks." + "reason": "Tokenizer special-token conflict and notebook custom-model initialization breakage are unrelated." }, { "left": "issue:43646", "right": "issue:44534", "accept": false, - "reason": "Different bugs: custom model initialization failure vs non-persistent buffers being filled with junk." + "reason": "Custom model initialization and junk non-persistent buffers are separate regressions." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Both are v5 loading regressions, but they target different repositories/models and don't look like the same defect." + "reason": "Different affected models and no clear shared bug." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Different issues: template compilation error vs GraniteMoEHybrid calling an invalid method." + "reason": "Template-node TypeError vs GraniteMoEHybrid invalid method call are unrelated." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Different symptom domains: nonsensical SigLIP2 outputs vs OmDet-Turbo processor/model input-size mismatch." + "reason": "Nonsensical outputs and processor image-size mismatch are different issues." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "MIMI encoder padding-mask batching bug is unrelated to DAC.from_latents missing STE mismatch." - }, - { - "left": "issue:43322", - "right": "issue:43540", - "accept": false, - "reason": "Different multimodal loading/runtime failures: Llava Next segfault vs Qwen3OmniMoe video ValueError." + "reason": "MIMI batching discrepancy and DAC from_latents mismatch are unrelated." }, { "left": "issue:43645", "right": "issue:44534", "accept": false, - "reason": "Custom model initialization in Jupyter is not the same bug as non-persistent buffer junk initialization." + "reason": "Notebook initialization breakage and buffer initialization junk are different bugs." }, { "left": "issue:45216", "right": "issue:45357", "accept": true, - "reason": "Same Qwen3.5 save_pretrained regression; 45357 is a narrower follow-up about incorrect visual encoder keys." + "reason": "Same Qwen3.5 save_pretrained regression; 45357 is a later symptom focused on incorrect visual encoder keys." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Llava Next segmentation fault and Qwen3OmniMoe video ValueError are unrelated." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "General loading noise is not the same as the Mistral regex patch crash." + "reason": "Loading noise and Mistral regex patch crash are different issues." }, { "left": "issue:44451", "right": "issue:45081", "accept": false, - "reason": "Different loading failures for different models and different tokenizer code paths." + "reason": "ScandiBERT load failure and Mistral tokenizer regex crash are unrelated." }, { "left": "issue:44265", "right": "issue:44610", "accept": false, - "reason": "torch.export export failure with torch_compilable_check is unrelated to processor image-size mismatch." + "reason": "torch.export failure and OmDet-Turbo processor size mismatch are not the same bug." }, { "left": "issue:43632", "right": "issue:44756", "accept": false, - "reason": "Transformer v5 initialization flag breakage is unrelated to mmap/OOM on Strix Halo." + "reason": "_is_hf_initialized flag regression and mmap/OOM on Strix Halo are unrelated." }, { "left": "issue:41553", "right": "issue:43582", "accept": false, - "reason": "Bad AutoTokenizer error message for Voxtral is unrelated to an AppleSilicon TypeError in caching_allocator_warmup." + "reason": "Voxtral tokenizer error message and Apple Silicon allocator TypeError are unrelated." }, { "left": "issue:44220", "right": "issue:44683", "accept": false, - "reason": "Different low-level problems: fbank feature extraction vs compiled flex_attention on newer torch." + "reason": "Audio feature extraction and compiled flex_attention torch 2.9 failure are different problems." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Both involve tied weights, but one is missing tying in model definition and the other is LoRA merge collapse with extended vocab." + "reason": "Both touch tied weights, but one is a missing tie and the other is a LoRA-merge collapse case." }, { "left": "issue:42915", "right": "issue:45084", "accept": false, - "reason": "Different bugs in different subsystems; FP8 model loading is not the same as template-node compilation failure." + "reason": "Qwen3Moe FP8 failure and template compilation TypeError are unrelated." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Both are Qwen vision/video regressions, but one is a broad video-input regression and the other is a specific still-image position-id scaling bug." + "reason": "This is only loosely related by Qwen VL/video-positioning terminology; one is a broad video-input regression, the other a specific still-image temporal position_ids scaling bug." }, { "left": "issue:44987", "right": "issue:45081", "accept": false, - "reason": "Unrelated: model loading failure for physical-intelligence/fast vs a Mistral tokenizer regex patch crash." + "reason": "Different model-loading/tokenizer problems with no evidence of the same root cause." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL load failure is unrelated to Qwen3-MoE sliding_window behavior." + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window issue are unrelated." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "High perplexity on gpt-oss-20b is unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "Perplexity anomaly and Whisper batch_decode skip_special_tokens handling are unrelated." } ] }, "evaluator_result": { - "accept": true, - "feedback": "Grounded overall. The duplicate anchor around 43644/44534 is well supported by the titles, and the other soft-edge rejections stay conservative. The Qwen3.5 save_pretrained pair is plausible as a duplicate-style cluster; nothing in the packet clearly contradicts that." + "accept": false, + "feedback": "The cluster summary is mostly grounded, but one accepted soft-edge verdict is too strong for the evidence in the packet. In particular, `issue:44479` vs `issue:45325` is only loosely related by Qwen VL/video-positioning terminology: one is a broad video-input regression across several Qwen models, while the other is a specific still-image temporal position_ids scaling bug. From titles alone, that should be treated as at most a plausible relation, not a true duplicate. The GGUF and Qwen3.5 save_pretrained pairs look more defensible, but this overreach means the cluster should be rejected." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -2664,29 +2727,29 @@ "issue:44062|issue:44987", "issue:42915|issue:44291", "issue:43317|issue:43856", - "issue:40444|issue:45325", "issue:36032|issue:43742", + "issue:40444|issue:45325", "issue:43632|issue:44534", "issue:43531|issue:44877", + "issue:42915|issue:44589", "issue:43421|issue:44568", "issue:43818|issue:45357", - "issue:42915|issue:44589", "issue:36032|issue:41720", "issue:43482|issue:45538", "issue:44829|issue:45479", "issue:44683|issue:45020", "issue:33666|issue:40444", "issue:43425|issue:44403", - "issue:44079|issue:44534", "issue:44403|issue:44991", + "issue:44079|issue:44534", "issue:35707|issue:40444", "issue:43377|issue:45491", "issue:43792|issue:44589", "issue:43482|issue:43525", "issue:44291|issue:44589", "issue:43749|issue:44079", - "issue:44079|issue:45072", "issue:43644|issue:44079", + "issue:44079|issue:45072", "issue:44265|issue:45072", "issue:43632|issue:45072", "issue:40444|issue:45381", @@ -2696,250 +2759,250 @@ "issue:41720|issue:45081", "issue:43881|issue:44488" ], - "prepared_review_unit_hash": "8f096225c2d87e5152847e03e92bee69df37f595bfe641c76ba687a059523ab3", + "prepared_review_unit_hash": "19e64b88efc6a87d379bc22ec4008054f502a5ddc56839e8fe883f51e0215850", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 36, - "serialized_chars": 38433, - "estimated_input_tokens": 9609, - "estimated_eval_tokens": 19474 + "serialized_chars": 38422, + "estimated_input_tokens": 9606, + "estimated_eval_tokens": 19468 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of distinct Transformers v5 / model-loading regressions across different models and subsystems. I don\u2019t see any strong duplicate pairs; the only clear commonality is broad version-compatibility breakage.", - "confidence": 0.31, - "canonical_issue_reason": "#45020 is the best umbrella issue: it is broad, open, and captures the recurring theme of recent Transformers versions breaking model loading / remote-code flows.", + "summary": "All soft pairs are false duplicates: they mostly share only a broad subsystem, model family, or Transformers-v5 regression theme, but each points to a different concrete failure mode. No single canonical issue stands out, and there is no PR evidence in this cluster.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue: the set spans unrelated bugs across tokenizers, loading, vision/video, device mapping, quantization, and training; the similarities are too broad to unify into one underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": "#45020 is the most representative because it\u2019s the broadest report in the set and matches the shared regression theme better than the model-specific bugs.", + "best_issue_reason": "No clear global best issue. A broad v5-regression report might look umbrella-like, but none of the issues cleanly subsumes the others as the same bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42175", "right": "issue:45542", "accept": false, - "reason": "Both mention backend/runtime install issues, but one is about missing TensorFlow in the backend and the other about tensorboard being installed without TF; different packaging bugs." + "reason": "Both concern TensorFlow-related install/backend behavior, but one is about missing TF in `pip install .[torch]` and the other about an undefined TF backend when only tensorboard is installed. Different packaging/config bugs." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "Both are Transformers v5 regressions, but the concrete problems differ: model loading with base_model_prefix vs corrupted non-persistent buffers." + "reason": "Both are Transformers v5 regressions, but one breaks loading models with `base_model_prefix` and the other corrupts non-persistent buffers. Different code paths and symptoms." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "Both concern tokenizer/model loading failures, but the titles point to different failure modes and no shared concrete code path is evident." + "reason": "The first is a tokenizer `AddedToken(..., special=...)` argument collision; the second is loading a specific model (`physical-intelligence/fast`) on newer Transformers. Too little evidence of the same defect." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "Both are load/init-related regressions, but one is FP8 config handling and the other is an unexpected _is_hf_initialized arg during init_empty_weights." + "reason": "One is a FineGrainedFP8Config failure on Qwen3MoE; the other is an `_is_hf_initialized` argument error with `init_empty_weights`. Different initialization/quantization bugs." }, { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "Different subsystems entirely: dequantized device_map loading vs Qwen3 MoE training memory usage." + "reason": "Both mention memory/offload concerns, but one is device_map dequantized loading on gpu+cpu offload and the other is inefficient Qwen3 MoE training memory use. Not the same bug." }, { - "left": "issue:40444", - "right": "issue:45325", + "left": "issue:36032", + "right": "issue:43742", "accept": false, - "reason": "Both are Qwen2.5-VL issues, but one is multi-image finetuning with IterableDataset and the other is rope/position-id scaling for still images." + "reason": "Tokenizer `add_special_tokens` conflict in T5Tokenizer versus a key error loading MobileLLM-125M. Same area only at a very high level." }, { - "left": "issue:36032", - "right": "issue:43742", + "left": "issue:40444", + "right": "issue:45325", "accept": false, - "reason": "Unrelated model-loading problems: a T5 tokenizer special-token conflict vs a key error loading MobileLLM-125M." + "reason": "Both involve Qwen2.5-VL, but one is iterable-dataset training with multiple images per prompt and the other is incorrect still-image temporal position scaling in `get_rope_index`. Different failures." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "Both are v5 regressions, but the underlying bugs are different: _is_hf_initialized handling vs non-persistent buffer initialization." + "reason": "Both are v5 regressions, but `_is_hf_initialized` handling and junk-filled non-persistent buffers are separate loading/initialization issues." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "Different models and different failures: Qwen3-MoE sliding_window behavior vs strict config loading for granite_speech." + "reason": "One is a Qwen3-MoE sliding-window issue; the other is strict config blocking `granite_speech`. No concrete overlap." }, { - "left": "issue:43421", - "right": "issue:44568", + "left": "issue:42915", + "right": "issue:44589", "accept": false, - "reason": "Both involve tokenizer special-token behavior, but one is runtime post-processor updates and the other is BOS/EOS not being added for a specific tokenizer." + "reason": "FineGrainedFP8Config failure is unrelated to the `Float8_e4m3fnStorage` lookup/type error. Same quantization theme, different bug." }, { - "left": "issue:43818", - "right": "issue:45357", + "left": "issue:43421", + "right": "issue:44568", "accept": false, - "reason": "Completely different model families and bugs: Video-LLaVA tower weights/temporal attention vs Qwen3.5 save_pretrained key names." + "reason": "Runtime post-processor updates for special tokens is not the same as `add_special_tokens=True` failing to add BOS/EOS for a specific tokenizer." }, { - "left": "issue:42915", - "right": "issue:44589", + "left": "issue:43818", + "right": "issue:45357", "accept": false, - "reason": "FP8-related loading problems, but one is FineGrainedFP8Config and the other is a missing Float8 storage object error." + "reason": "Video-LLaVA missing temporal attention/weight sharing is unrelated to Qwen3.5 `save_pretrained` writing wrong visual encoder keys." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Different failures: tokenizer special-token conflict vs Qwen3 auto device mapping cudaErrorAssert." + "reason": "Tokenizer load conflict and Qwen3 auto device mapping CUDA assert are unrelated failures." }, { "left": "issue:43482", "right": "issue:45538", "accept": false, - "reason": "No concrete overlap: Qwen2.5-GGUF loading under v5 vs CLIPTokenizer model_max_length behavior." + "reason": "Qwen2.5-GGUF loading with Transformers v5 and CLIPTokenizer `model_max_length` being 10**30 are different tokenizer/model-loading problems." }, { "left": "issue:44829", "right": "issue:45479", "accept": false, - "reason": "Both are classification-training regressions, but one is flash_attention_3 degenerate training and the other is num_labels=1 zero-loss behavior." + "reason": "Both concern degenerate classification training, but one is tied to `flash_attention_3` and the other to `num_labels=1`/`problem_type`. Not the same concrete issue." }, { "left": "issue:44683", "right": "issue:45020", "accept": false, - "reason": "Different code paths: compiled flex_attention on torch>=2.9 vs remote_code model-loading regressions in recent Transformers." + "reason": "Compiled flex_attention on torch>=2.9 and remote_code loading regressions are unrelated." }, { "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 finetuning with multiple images per prompt." + "reason": "Both are Qwen VL training issues, but multi-GPU training and iterable-dataset multi-image failures are distinct problems." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Unrelated: Torch 2.10 compatibility vs a noisy model-loading message." + "reason": "Torch 2.10 compatibility and generic loading noise are not the same bug." }, { - "left": "issue:44079", - "right": "issue:44534", + "left": "issue:44403", + "right": "issue:44991", "accept": false, - "reason": "Different bugs: ModelOutput key assignment vs non-persistent buffer initialization." + "reason": "Unnecessary loading noise and tokenizer load failure for EMBEDDIA/est-roberta are unrelated." }, { - "left": "issue:44403", - "right": "issue:44991", + "left": "issue:44079", + "right": "issue:44534", "accept": false, - "reason": "No shared underlying bug; one is loading noise, the other is tokenizer loading failure for a specific model." + "reason": "ModelOutput key assignment with `None` versus non-persistent buffer corruption are separate internals issues." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Different areas entirely: progressive generation with inputs_embeds/past_key_values vs Qwen2.5-VL iterable-dataset finetuning." + "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is unrelated to Qwen2.5-VL iterable-dataset multi-image training." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "Both involve batching/padding, but the concrete bugs differ: MIMI padding mask support vs Gemma3 NaNs from sliding-window all-padding cases." + "reason": "Both involve batched vs single-input discrepancies, but MIMI encoder padding-mask support and Gemma3 sliding-window all-padding NaNs are different models and mechanisms." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Different model/runtime failures: Whisper loading vs Float8 storage resolution." + "reason": "Whisper-large-v2 not running and `Float8_e4m3fnStorage` lookup failure are unrelated." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Both are model-loading regressions, but one is GGUF v5 loading and the other is a missing pad_token_id on Llama4Config." + "reason": "GGUF loading under Transformers v5 and missing `pad_token_id` on `Llama4Config` are distinct loading/config issues." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "Different failures: init_empty_weights/_is_hf_initialized vs Float8 storage lookup." + "reason": "Unexpected `_is_hf_initialized` in init-empty-weights and missing float8 storage are separate v5-related failures." }, { "left": "issue:43749", "right": "issue:44079", "accept": false, - "reason": "No overlap: FSDP CPU RAM efficient loading vs ModelOutput key assignment." + "reason": "Broken FSDP CPU RAM efficient loading and misassigned ModelOutput keys are unrelated." }, { - "left": "issue:44079", - "right": "issue:45072", + "left": "issue:43644", + "right": "issue:44079", "accept": false, - "reason": "Different bugs: ModelOutput key handling vs bfloat16 dtype mismatches in CI." + "reason": "Non-persistent buffer junk and ModelOutput key assignment bugs affect different code paths." }, { - "left": "issue:43644", - "right": "issue:44079", + "left": "issue:44079", + "right": "issue:45072", "accept": false, - "reason": "Non-persistent buffer junk vs ModelOutput key reassignment are unrelated." + "reason": "ModelOutput key handling and dtype mismatches in SwitchTransformers/TimmWrapperModel are different issues." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "Different subsystems: torch.export with torch_compilable_check vs bfloat16 dtype mismatch in inference." + "reason": "`torch.export.export` failure via `torch_compilable_check` is unrelated to bfloat16 dtype mismatches." }, { "left": "issue:43632", "right": "issue:45072", "accept": false, - "reason": "No shared concrete code path: _is_hf_initialized flag breakage vs dtype mismatches." + "reason": "`_is_hf_initialized` breakage and dtype mismatch CI failures are not the same bug." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "Both are Qwen2.5-VL vision issues, but one is multi-image finetuning and the other is wrong video vision_position_ids." + "reason": "Both are Qwen2.5-VL vision issues, but one is multi-image iterable training failure and the other is wrong video vision_position_ids. Different concrete paths." }, { "left": "issue:45325", "right": "issue:45381", "accept": false, - "reason": "Related area, but not the same bug: still-image temporal position_id scaling vs video input vision_position_ids." + "reason": "Both concern Qwen2.5-VL position IDs, but one is still-image temporal scaling and the other is video input position IDs. Related area, not the same defect." }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Both are v5 loading regressions, but the specific failures are different: base_model_prefix loading vs custom model initialization." + "reason": "`base_model_prefix` loading regression and custom model initialization breakage are separate Transformers v5 issues." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Different model-specific regressions: Qwen3-MoE sliding_window vs init_empty_weights unexpected argument." + "reason": "Qwen3-MoE sliding-window behavior and `_is_hf_initialized` init-empty-weights errors do not share the same code path." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "No overlap: Qwen3 auto device mapping crash vs Mistral tokenizer patch crashing on missing backend_tokenizer." + "reason": "Qwen3 auto device mapping CUDA assert and Mistral tokenizer regex patch crash are unrelated." }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "Different model-loading failures for different models; no evidence they share the same root cause." + "reason": "Loading `glm-4v-9b` and loading `cjvt/sleng-bert` fail for different models and likely different causes." } ] }, @@ -3014,8 +3077,8 @@ ], "soft_pairs": [ "issue:43454|issue:44466", - "issue:43994|issue:44898", "issue:42175|issue:43296", + "issue:43994|issue:44898", "issue:44387|issue:44617", "issue:43828|issue:44683", "issue:42915|issue:43482", @@ -3025,8 +3088,8 @@ "issue:43792|issue:43957", "issue:43632|issue:45563", "issue:43317|issue:43873", - "issue:43881|issue:44991", "issue:43606|issue:44265", + "issue:43881|issue:44991", "issue:34634|issue:35707", "issue:44186|issue:44265", "issue:44291|issue:44877", @@ -3034,210 +3097,211 @@ "issue:43377|issue:44898", "issue:35707|issue:41720", "issue:44610|issue:45325", - "issue:43454|issue:43550", "issue:43454|issue:45072", + "issue:43454|issue:43550", "issue:43377|issue:44610", "issue:39692|issue:40444", "issue:43749|issue:44534", "issue:44466|issue:45357", "issue:43296|issue:44991", "issue:43454|issue:43720", - "issue:41720|issue:43322", "issue:43296|issue:44877", + "issue:41720|issue:43322", "issue:41720|issue:44977", "issue:43792|issue:44683", - "issue:43404|issue:45357" + "issue:43404|issue:45357", + "issue:44186|issue:45072" ], - "prepared_review_unit_hash": "342fec754bd49961bed5fab1145ed8d84801baaa5d8fba868601ef72bc5771c4", + "prepared_review_unit_hash": "857c036006fe2bc187070b5914c40e77798206c62427673b3c6d7a846a03fe01", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37863, - "estimated_input_tokens": 9466, - "estimated_eval_tokens": 19188 + "soft_pair_count": 35, + "serialized_chars": 38166, + "estimated_input_tokens": 9542, + "estimated_eval_tokens": 19340 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items are mostly a heterogeneous mix of model-specific regression reports. A few pairs share broad surface area (offloading, tied weights, loading), but none look like the same concrete bug closely enough to merge.", - "confidence": 0.62, - "canonical_issue_reason": "Issue 43873 is the broadest and least model-specific report in the set; it names the core offloading/quantization symptom directly and has active references, making it the best available hub for this mostly mixed cluster.", + "summary": "These items are mostly unrelated model-loading, quantization, and generation bugs. I found no true duplicate pairs; the soft matches are only broad subsystem similarity, not the same concrete bug or change.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43873 is the broadest still-open report here and best captures the shared offloading/quantization theme, but the cluster is too heterogeneous for a strong duplicate canonical.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43873 is the most representative single issue for triage because it is broad, open, and framed around the underlying runtime problem rather than a narrow model-specific manifestation.", + "best_issue_reason": "Issue 43873 is the best global anchor because it is broad, active, and has the clearest cross-cutting loading/offload scope among the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Both involve lm_head/tied-weights concerns, but one is a model-specific tie failure and the other is a serialization inconsistency; not the same concrete bug." + "reason": "Both mention tied lm_head weights, but one is a model-specific tie bug and the other is a device-dependent serialization inconsistency; not the same concrete failure." }, { - "left": "issue:43994", - "right": "issue:44898", + "left": "issue:42175", + "right": "issue:43296", "accept": false, - "reason": "Both are vision-model correctness issues, but they affect different models and different failure modes." + "reason": "Completely different problems: missing TensorFlow backend in an install extra vs a PaddleOCR-VL loading failure in vLLM." }, { - "left": "issue:42175", - "right": "issue:43296", + "left": "issue:43994", + "right": "issue:44898", "accept": false, - "reason": "Unrelated packaging/backend installation issue versus an audio processor sampling-rate bug." + "reason": "Both are vision-input mismatches, but they involve different models and different root causes; too broad to treat as one bug." }, { "left": "issue:44387", "right": "issue:44617", "accept": false, - "reason": "Both mention memory/runtime trouble, but they are different models and different root causes." + "reason": "Both end in OOM, but one is a Transformers int4 memory regression and the other is Sam3Video model memory usage; not the same issue." }, { "left": "issue:43828", "right": "issue:44683", "accept": false, - "reason": "Autocast dtype mismatch and compiled flex_attention failure are different code-path regressions." + "reason": "Dtype-mismatch under autocast and compiled flex_attention failure on new torch versions are distinct code paths." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "Different models and different failures: FineGrainedFP8Config support versus GGUF/tokenizer loading on v5." + "reason": "Different models and different breakages: FP8 config failure vs GGUF loading failure under v5." }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "Image processor label reduction bug is unrelated to the Molmo tied-weights attribute error." + "reason": "One is a BeitImageProcessorFast label-reduction bug; the other is a missing all_tied_weights_keys attribute on MolmoForCausalLM." }, { "left": "issue:43701", "right": "issue:45325", "accept": false, - "reason": "Checkpoint resume key mismatch is unrelated to Qwen2.5-VL rope/position-id scaling." + "reason": "Checkpoint key mismatch and Qwen2.5-VL rope-index scaling are unrelated generation/loading bugs." }, { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Audio chat-template sampling-rate default and GraniteSpeech PEFT checkpoint loading are unrelated." + "reason": "Audio chat-template sampling-rate defaults and GraniteSpeech PEFT checkpoint loading are different subsystems and bugs." }, { "left": "issue:43792", "right": "issue:43957", "accept": false, - "reason": "Different model-loading failures with no shared code-path indicated." + "reason": "Whisper runtime failure vs meta-device loading regressions in v5 are separate model-loading problems." }, { "left": "issue:43632", "right": "issue:45563", "accept": false, - "reason": "_is_hf_initialized flag breakage is unrelated to the stale warning for num_return_sequences." + "reason": "_is_hf_initialized flag breakage and a stale warning in paged generate are unrelated." }, { "left": "issue:43317", "right": "issue:43873", "accept": false, - "reason": "Both involve offloading/quantization broadly, but one is dequantized model loading on cpu/gpu offload and the other is a general offloading regression; not clearly the same bug." + "reason": "Both involve offloading, but one is a dequantized model load failure with device_map=auto and the other is a broader quantization/offload behavior report." }, { - "left": "issue:43881", - "right": "issue:44991", + "left": "issue:43606", + "right": "issue:44265", "accept": false, - "reason": "Different models and different loading layers: model load failure versus tokenizer loading regression." + "reason": "CPU offload device mismatch and torch.export failures with torch_compilable_check are not the same bug." }, { - "left": "issue:43606", - "right": "issue:44265", + "left": "issue:43881", + "right": "issue:44991", "accept": false, - "reason": "CPU-offload device mismatch and torch.export failure via torch_compilable_check are unrelated." + "reason": "GLM-4V model loading and tokenizer loading for est-roberta are unrelated." }, { "left": "issue:34634", "right": "issue:35707", "accept": false, - "reason": "Bark voice_preset and progressive generation with inputs_embeds/past_key_values are unrelated features." + "reason": "Bark voice_preset and progressive generation with inputs_embeds/past_key_values are different features with different failures." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "Tokenizer padding/NER crash is unrelated to export failures around torch_compilable_check." + "reason": "Tokenizer crashes on NER/padding vs export failures from torch_compilable_check are unrelated." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "Unexpected _is_hf_initialized argument and strict config rejection are different loading issues." + "reason": "Unexpected init_empty_weights argument and strict config rejection are different loading-time issues." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "Device mismatch on CPU offload is unrelated to DAC latent decoding mismatch." + "reason": "Both are model bugs, but one is offload/device mismatch and the other is DAC.from_latents mismatch; not the same code path." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "Missing padding-mask support in MIMI encoder is unrelated to Perceiver resolution handling." + "reason": "MIMI batch-vs-single padding-mask behavior and Perceiver resolution handling are different bugs." }, { "left": "issue:35707", "right": "issue:41720", "accept": false, - "reason": "Progressive generation with cache inputs and Qwen3 device-map CUDA assertion are different bugs." + "reason": "Progressive generation with cached embeddings and Qwen3 auto device-map CUDA asserts are unrelated." }, { "left": "issue:44610", "right": "issue:45325", "accept": false, - "reason": "Processor output size mismatch and Qwen2.5-VL temporal rope scaling are unrelated." + "reason": "Processor output resolution mismatch and Qwen2.5-VL rope-index scaling are different model-specific issues." }, { "left": "issue:43454", - "right": "issue:43550", + "right": "issue:45072", "accept": false, - "reason": "Tied lm_head weights and torch.compile SDPA failure do not share a concrete code-path." + "reason": "lm_head weight tying and bfloat16 dtype mismatches are unrelated." }, { "left": "issue:43454", - "right": "issue:45072", + "right": "issue:43550", "accept": false, - "reason": "Tied-weight generation bug and bfloat16 dtype mismatch are unrelated." + "reason": "Weight-tying bug vs torch.compile/SDPA failure; no shared concrete bug." }, { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "Missing padding-mask support in MIMI and OmDet-Turbo processor resolution mismatch are unrelated." + "reason": "MIMI padding-mask correctness and OmDet-Turbo image-size mismatch are different subsystems." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "SigLIP2 doc/example errors and Qwen2.5-VL finetuning failures are different tasks and failures." + "reason": "Documentation/example errors for SigLIP2 and multi-image finetuning failures in Qwen2.5-VL are separate reports." }, { "left": "issue:43749", "right": "issue:44534", "accept": false, - "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk in v5 are different regressions." + "reason": "FSDP CPU RAM-efficient loading is not the same as non-persistent buffer initialization junk." }, { "left": "issue:44466", "right": "issue:45357", "accept": false, - "reason": "Both touch v5 save/serialization, but one is lm_head tied-weight serialization and the other is incorrect visual encoder keys for Qwen3.5." + "reason": "Both touch serialization/save_pretrained behavior, but one is lm_head tying on device and the other is incorrect visual encoder keys in Qwen3.5." }, { "left": "issue:43296", @@ -3249,37 +3313,43 @@ "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "lm_head tie-weight issue is unrelated to packed-weight unpacking during accelerate loading." + "reason": "Weight tying in a model head is unrelated to BitNet packed-weight unpacking during accelerate loading." }, { - "left": "issue:41720", - "right": "issue:43322", + "left": "issue:43296", + "right": "issue:44877", "accept": false, - "reason": "Qwen3 device-map CUDA assert and Llava Next segmentation fault are different failures." + "reason": "PaddleOCR-VL load failure and strict granite_speech config rejection are different failures." }, { - "left": "issue:43296", - "right": "issue:44877", + "left": "issue:41720", + "right": "issue:43322", "accept": false, - "reason": "PaddleOCR-VL load failure and strict granite_speech config loading are unrelated." + "reason": "Auto device-map CUDA assert and Llava Next segmentation fault are both loading failures but not the same bug." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "CUDA assert on auto device mapping and flash-attention generation issues in Qwen3.5 are different bugs." + "reason": "Qwen3 auto device mapping crash and Qwen3.5 flash-attention generation bug are distinct issues." }, { "left": "issue:43792", "right": "issue:44683", "accept": false, - "reason": "Whisper runtime failure and compiled flex_attention regression are unrelated." + "reason": "Whisper runtime failure and compiled flex_attention failure on torch>=2.9 are unrelated." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "lm_head weight tying in Mistral3 and incorrect visual encoder keys in Qwen3.5 save_pretrained are unrelated." + "reason": "Mistral3 lm_head tying and Qwen3.5 visual encoder key serialization are different model-specific regressions." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer/padding crashes and bfloat16 dtype mismatches do not share the same underlying bug." } ] }, @@ -3303,8 +3373,10 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-366", "nodes": [ + "issue:30064", "issue:33357", "issue:34567", + "issue:34634", "issue:35707", "issue:36010", "issue:36331", @@ -3346,14 +3418,12 @@ "issue:44918", "issue:44991", "issue:45059", - "issue:45072", "issue:45161", "issue:45406", "issue:45464" ], "soft_pairs": [ "issue:44423|issue:45406", - "issue:44186|issue:45072", "issue:34567|issue:40990", "issue:43526|issue:43881", "issue:43873|issue:44756", @@ -3361,8 +3431,8 @@ "issue:42617|issue:42915", "issue:43550|issue:43720", "issue:44610|issue:44898", - "issue:44186|issue:44610", "issue:44877|issue:44991", + "issue:44186|issue:44610", "issue:44479|issue:44918", "issue:36331|issue:42915", "issue:36331|issue:42617", @@ -3374,8 +3444,8 @@ "issue:44292|issue:44479", "issue:41720|issue:45161", "issue:43531|issue:43572", - "issue:43388|issue:45059", "issue:44186|issue:44361", + "issue:43388|issue:45059", "issue:43540|issue:43782", "issue:41720|issue:43873", "issue:42915|issue:43322", @@ -3385,162 +3455,157 @@ "issue:44514|issue:44545", "issue:33357|issue:35707", "issue:43606|issue:43720", - "issue:41720|issue:43746" + "issue:41720|issue:43746", + "issue:30064|issue:34634" ], - "prepared_review_unit_hash": "0f0ccaa5e8e19f2bc899c5dad6e345d49b18f7f08077dae38f51243b7e75ac64", + "prepared_review_unit_hash": "954bcc5436b28a3fb5fa146cd92f443b4a084ae14e781a23b54219298681e987", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37216, - "estimated_input_tokens": 9304, - "estimated_eval_tokens": 18864 + "serialized_chars": 37728, + "estimated_input_tokens": 9432, + "estimated_eval_tokens": 19120 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This packet is largely heterogeneous. I accept only one soft edge: the two Qwen2_5_VLProcessor.apply_chat_template reports describe the same underlying batched-input crash with padding=False, even though their titles differ slightly in formatting. The rest look like distinct bugs, and I keep those rejections conservative.", - "confidence": 0.87, - "canonical_issue_reason": "issue:44545 is the clearest canonical issue for the accepted duplicate pair: it describes the same batched-input apply_chat_template crash as issue:44514, with a slightly cleaner normalized title and a bit more inbound discussion signal.", + "summary": "This set is mostly heterogeneous single-bug issues with only one clear duplicate pair: Qwen2_5_VLProcessor.apply_chat_template crashes on batched input when padding=False (44514/44545). The rest look like related-sounding but distinct bugs, so I rejected the soft edges.", + "confidence": 0.86, + "canonical_issue_reason": "issue:44545 is the best canonical anchor here because it is the clearest, most concise representation of the only exact duplicate pair (with 44514) and has the stronger reference signal.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44545 is the best issue to keep as the representative node for this duplicate pair; it is the cleaner semantically equivalent entry and slightly better connected by references.", + "best_issue_reason": "issue:44545 is the strongest standalone issue in this set: it is specific, reproducible, and clearly duplicated by 44514, making it the best representative of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44423", "right": "issue:45406", "accept": false, - "reason": "Both are serve/runtime crashes, but they involve different models and different failure points." - }, - { - "left": "issue:44186", - "right": "issue:45072", - "accept": false, - "reason": "Tokenizer padding/truncation crash and bfloat16 dtype mismatch are unrelated bugs." + "reason": "Both are serve/runtime crashes in multimodal contexts, but they involve different processors and failure modes ('str' .to vs missing _tokenizer)." }, { "left": "issue:34567", "right": "issue:40990", "accept": false, - "reason": "TrainerState token counting and GPT-OSS perplexity are different symptoms and code paths." + "reason": "Trainer token counting and perplexity on GPT-OSS are unrelated bugs." }, { "left": "issue:43526", "right": "issue:43881", "accept": false, - "reason": "BEiT label reduction and glm-4v loading failure are unrelated." + "reason": "BeitImageProcessorFast label reduction and GLM-4V loading failure are different code paths and symptoms." }, { "left": "issue:43873", "right": "issue:44756", "accept": false, - "reason": "Both mention offloading/memory, but one is quantization offload behavior and the other is mmap OOM on Strix Halo." + "reason": "Both touch offloading/memory, but one is quantization offload behavior and the other is mmap/OOM on Strix Halo; not the same bug." }, { "left": "issue:43577", "right": "issue:43708", "accept": false, - "reason": "Model dtype loading and Trainer resume/max_steps calculation are separate issues." + "reason": "BLIP2 dtype propagation and Trainer resume max_steps calculation are unrelated." }, { "left": "issue:42617", "right": "issue:42915", "accept": false, - "reason": "3d_parallel.py execution failure and Qwen3Moe FP8 failure are distinct." + "reason": "A 3d_parallel.py failure and a FineGrainedFP8Config failure are different model/runtime problems." }, { "left": "issue:43550", "right": "issue:43720", "accept": false, - "reason": "torch.compile+SDPA failure and packed-weights loading failure are not the same code-path bug." + "reason": "Both involve model loading/execution infrastructure, but the concrete failures differ: torch.compile+SDPA vs packed weights not unpacked during accelerate loading." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Both are image-size related, but they are different models with different preprocessing expectations." + "reason": "These are both image-size/resolution issues, but on different models and with different expected-processing mismatches." }, { - "left": "issue:44186", - "right": "issue:44610", + "left": "issue:44877", + "right": "issue:44991", "accept": false, - "reason": "Tokenizer crash on NER inputs is unrelated to OmDet-Turbo producing the wrong image size." + "reason": "Strict config loading for granite_speech and tokenizer loading for EMBEDDIA/est-roberta are unrelated." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:44186", + "right": "issue:44610", "accept": false, - "reason": "Strict config loading and tokenizer loading for a specific model are different issues." + "reason": "Tokenizer crashes on NER/padding vs processor output size mismatch are different bugs." }, { "left": "issue:44479", "right": "issue:44918", "accept": false, - "reason": "One is a video-input regression across Qwen VL models; the other is a TRL SFT unpacking failure for Qwen3.5 embeddings." + "reason": "Both mention Qwen-family models, but one is a video-input regression and the other is an embeddings/trainer failure." }, { "left": "issue:36331", "right": "issue:42915", "accept": false, - "reason": "CustomTrainer compute_loss signature mismatch and Qwen3Moe FP8 loading are unrelated." + "reason": "Unexpected compute_loss kwargs and FineGrainedFP8 model failure are unrelated." }, { "left": "issue:36331", "right": "issue:42617", "accept": false, - "reason": "Trainer API mismatch and 3d_parallel.py runtime failure are unrelated." + "reason": "Trainer loss signature mismatch is unrelated to the 3d_parallel.py failure." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "CamemBERT masked-LM regression and Qwen3.5 recurrent-state cache bug are different." + "reason": "CamemBERT masked-LM predictions and Qwen modular recurrent-state reset are different model-specific bugs." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Gemma3TextConfig API mismatch and GPTNeoX rotary_pct reload bug are unrelated config issues." + "reason": "Both are config/serialization regressions, but they concern different fields and models; not the same underlying issue." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Different failure modes: SDPA/torch.compile vs CPU offload device mismatch." + "reason": "torch.compile+SDPA on Bamba and CPU-offload device mismatch on Bark are distinct failures." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "Qwen3Moe FP8 failure and dequantized-model offload loading are not the same bug." + "reason": "Qwen3Moe FineGrainedFP8 loading and dequantized model offload loading are both loading issues, but not the same code-path problem." }, { "left": "issue:41720", "right": "issue:42617", "accept": false, - "reason": "CUDA assert on auto device mapping and 3d_parallel.py not running are different problems." + "reason": "Qwen3 auto device mapping cudaErrorAssert and 3d_parallel.py failure are different problems." }, { "left": "issue:44292", "right": "issue:44479", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime issue and a video-input regression across Qwen VL models are distinct." + "reason": "Qwen NVFP4 runtime error and Qwen video-input regression are distinct." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "Both are Qwen/MoE-related, but one is an auto device-mapping CUDA assert and the other is tensor-parallel not working." + "reason": "Auto device mapping cudaErrorAssert on A800 is unrelated to GPT-OSS MoE TP failure." }, { "left": "issue:43531", @@ -3549,85 +3614,85 @@ "reason": "Sliding-window behavior in Qwen3-MoE and missing pad_token_idx in StableLmConfig are unrelated." }, { - "left": "issue:43388", - "right": "issue:45059", + "left": "issue:44186", + "right": "issue:44361", "accept": false, - "reason": "Metric gathering label truncation and SAM3 text/bbox behavior are different data-processing bugs." + "reason": "Different tokenizer bugs with different models and failure mechanisms." }, { - "left": "issue:44186", - "right": "issue:44361", + "left": "issue:43388", + "right": "issue:45059", "accept": false, - "reason": "Both are tokenizer-related crashes, but they involve different tokenizers, tasks, and failure mechanisms." + "reason": "gather_for_metrics label dropping and SAM3 PCS text/box behavior are unrelated." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Video-input processing failure and weight_only=True from_pretrained error are not the same bug." + "reason": "Qwen3OmniMoe video processing ValueError and Qwen3VL weight_only load error are separate multimodal issues." }, { "left": "issue:41720", "right": "issue:43873", "accept": false, - "reason": "CUDA assert on device mapping and quantization offloading behavior are different issues." - }, - { - "left": "issue:42915", - "right": "issue:43322", - "accept": false, - "reason": "Qwen3Moe FP8 failure and Llava Next segmentation fault are unrelated model-loading failures." + "reason": "Both mention device/offload behavior, but one is a cudaErrorAssert during auto mapping and the other is quantization offload behavior; not the same bug." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "GenerationMixin import error and 3d_parallel.py startup failure are unrelated." + "reason": "GenerationMixin import failure and 3d_parallel.py failure are unrelated." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Serve crash with Gemma4Processor and streaming chat/completions failure on Qwen3.5 are different model-specific issues." + "reason": "Gemma4Processor missing _tokenizer and Qwen3.5 streaming API failure are different serve bugs." }, { "left": "issue:36331", "right": "issue:41720", "accept": false, - "reason": "Trainer compute_loss API mismatch and Qwen3 device-mapping CUDA assert are unrelated." + "reason": "Trainer compute_loss signature mismatch is unrelated to Qwen3 device-mapping crashes." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "Same underlying Qwen2_5_VLProcessor.apply_chat_template batched-input crash with padding=False; the titles differ only in formatting, not substance." + "reason": "These are the same bug reported twice: identical Qwen2_5_VLProcessor.apply_chat_template batched-input crash with padding=False." }, { "left": "issue:33357", "right": "issue:35707", "accept": false, - "reason": "MacOS bus error with CLIP and progressive generation with inputs_embeds/past_key_values are unrelated." + "reason": "MacOS bus error on CLIP and progressive generation with inputs_embeds/past_key_values are different issues." }, { "left": "issue:43606", "right": "issue:43720", "accept": false, - "reason": "CPU offload device mismatch and packed-weights unpacking during accelerate loading are different bugs." + "reason": "Bark CPU-offload device mismatch and BitNet packed-weight unpacking failure are not the same concrete bug." }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Qwen3 auto device mapping CUDA assert and GraniteSpeech PEFT local-checkpoint loading are unrelated." + "reason": "Qwen3 device-mapping crash and GraniteSpeech PEFT checkpoint loading are unrelated." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Void segmentation-map processing and BarkProcessor voice_preset failure are unrelated model/processor bugs." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The duplicate call on issue:44514 \u2194 issue:44545 is well supported, and the other rejections are conservative. However, the writeup is a bit too strong in places: the two titles are not exactly identical, only semantically the same, and the claim that this is the 'only true duplicate pair' goes beyond what the packet itself establishes. Please soften those canonicalization/uniqueness assertions." + "accept": true, + "feedback": "Grounded and conservative. The only clearly duplicated pair in the packet is 44514/44545, and choosing 44545 as canonical/best issue is supported by the stronger inbound-reference signal. The rejected soft edges are phrased cautiously and do not overclaim equivalence based on the packet." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -3645,7 +3710,6 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-366", "nodes": [ - "issue:30064", "issue:34567", "issue:34634", "issue:35707", @@ -3695,7 +3759,6 @@ "issue:45479" ], "soft_pairs": [ - "issue:30064|issue:34634", "issue:43550|issue:43653", "issue:44492|issue:44509", "issue:44368|issue:45357", @@ -3714,243 +3777,237 @@ "issue:44062|issue:44451", "issue:44918|issue:44936", "issue:44987|issue:45042", - "issue:39692|issue:44610", - "issue:43994|issue:44671", "issue:43531|issue:44589", + "issue:43994|issue:44671", + "issue:39692|issue:44610", "issue:43421|issue:44186", "issue:45081|issue:45356", "issue:39692|issue:43873", - "issue:43746|issue:45381", - "issue:45183|issue:45479", "issue:43531|issue:43645", + "issue:45183|issue:45479", + "issue:43746|issue:45381", "issue:43596|issue:44220", "issue:43746|issue:43873", "issue:40444|issue:43746" ], - "prepared_review_unit_hash": "44ffeea966a396c6b91d11c1da6b8a2e51031f0331b28421e29232d33a44315d", + "prepared_review_unit_hash": "bc1a8b98bda57bb1f946572eb9d852ea8e5a9166ca7093e552e8a93d26760fcb", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37268, - "estimated_input_tokens": 9317, - "estimated_eval_tokens": 18890 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36412, + "estimated_input_tokens": 9103, + "estimated_eval_tokens": 18462 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of unrelated issues across tokenizers, vision processors, training, and model loading. Only a small tokenizer-regex/offline subset looks like genuine duplication.", - "confidence": 0.89, - "canonical_issue_reason": "No single canonical issue fits the whole cluster; the overlap is too weak and mostly limited to a small tokenizer-regex/offline subset.", - "canonical_pr_reason": "No canonical PR candidate is present in the data.", - "best_issue_reason": "If forced to pick one representative, issue 45081 is the closest anchor for the tokenizer-regex/fix_mistral_regex subset, but it is not representative of the cluster as a whole.", - "best_pr_reason": "No PR candidate is present.", + "summary": "This cluster is mostly unrelated issues from the same repo. The offline-tokenizer pair (42886 \u2194 44843) is thematically close, but with only title-level evidence it is not strong enough to call them the same bug or a mergeable duplicate, so it should be rejected as a soft edge.", + "confidence": 0.74, + "canonical_issue_reason": "issue:44843 is still the best anchor because it names a concrete suspected root cause (`model_info()` in `_patch_mistral_regex`) and directly describes the offline-loading failure mode.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44843 is the best global issue candidate because it is the most specific, actionable, and likely to subsume the broader offline tokenizer regression space among the provided issues.", + "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:30064", - "right": "issue:34634", - "accept": false, - "reason": "Different bugs in different processors: void segmentation maps vs Bark voice_preset handling." - }, { "left": "issue:43550", "right": "issue:43653", "accept": false, - "reason": "Unrelated failures: torch.compile/SDPA model bug vs tokenizer special-token registration." + "reason": "Different components and failure modes: torch.compile/SDPA model execution vs tokenizer special-token registration." }, { "left": "issue:44492", "right": "issue:44509", "accept": false, - "reason": "Both are docs/wording issues, but they concern different doc sets and different changes." + "reason": "Both are docs/wording-related, but they address different content and different changes." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "Different Qwen3.5 config issues: tie_word_embeddings warning vs visual encoder key saving." + "reason": "Unrelated Qwen3.5 config warning vs incorrect visual encoder key saving; different bug paths." }, { "left": "issue:44625", "right": "issue:45479", "accept": false, - "reason": "Both involve classification configs, but one is Qwen3.5 num_labels propagation and the other is a generic zero-loss problem_type bug." + "reason": "One is config propagation in Qwen3.5, the other is a generic classification loss degeneracy across models." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "Device mismatch with CPU offload is unrelated to BigBirdTokenizer special-token registration." + "reason": "CPU offload device mismatch is unrelated to a tokenizer special-token decode bug." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Both touch special tokens, but one is runtime post-processor refresh and the other is save/load token ID loss." + "reason": "Both involve token IDs/special tokens, but one is runtime post-processor updates and the other is save/load losing custom IDs; not the same bug." }, { "left": "issue:44898", "right": "issue:45072", "accept": false, - "reason": "Different inference bugs: image resolution handling vs bfloat16 dtype mismatches." + "reason": "Different model families and different problems: image resolution handling vs bfloat16 dtype mismatch." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Different Qwen-VL failures: from_pretrained weight_only handling vs wrong vision_position_ids for video input." + "reason": "Both are Qwen VL issues, but one is `from_pretrained(weight_only=True)` loading and the other is video vision_position_ids; not the same code path." }, { "left": "issue:34567", "right": "issue:35707", "accept": false, - "reason": "Training-state token counting and progressive generation are unrelated." + "reason": "Trainer state token counting is unrelated to progressive generation with inputs_embeds/past_key_values." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "Different model families and failures: BarkProcessor voice_preset vs Qwen3 auto device mapping/cuda assert." + "reason": "Bark voice_preset and Qwen3 auto device mapping crashes are unrelated." }, { "left": "issue:43381", "right": "issue:43844", "accept": false, - "reason": "Eval-mode gradient checkpointing vs ZeRO-3 gradient blow-up are distinct training behaviors." + "reason": "Gradient checkpointing in eval mode is a usage/validation issue, while the other is a ZeRO-3 gradient abnormality during training." }, { "left": "issue:42886", "right": "issue:44843", - "accept": true, - "reason": "Same tokenizer-loading offline failure family; 44843 identifies the unguarded model_info() call in the Mistral regex patch path, which matches the offline cache symptom." + "accept": false, + "reason": "Title-level similarity around HF_HUB_OFFLINE/tokenizer loading is not enough to confirm the same root cause; the pair is plausible but too weak to merge as a duplicate." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "Different config issues: patchtsmixer post_init policy vs Qwen3.5 num_labels propagation." + "reason": "Optional `post_init` policy for patchtsmixer is unrelated to Qwen3.5 num_labels propagation." }, { "left": "issue:43421", "right": "issue:43653", "accept": false, - "reason": "Both mention tokens, but one is post-processor refresh on runtime edits and the other is a tokenizer decode/special-token registration bug." + "reason": "Special-token runtime updates and BigBird mask-token registration are different tokenizer bugs." }, { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "Different tokenizer-loading regressions for different models and likely different root causes." + "reason": "Both are tokenizer-loading regressions, but they affect different models and likely different root causes." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "AddedToken constructor error is not the same as a model-specific tokenizer load failure." + "reason": "`AddedToken` argument collision is not the same as model-specific tokenizer loading failure." }, { "left": "issue:44918", "right": "issue:44936", "accept": false, - "reason": "Different trainer failures: unpacking Qwen3.5 embeddings vs evaluate-after-train state handling." + "reason": "TRL SFT unpacking with inputs_embeds is unrelated to trainer.evaluate() failing after train()." }, { "left": "issue:44987", "right": "issue:45042", "accept": false, - "reason": "Loading a specific model vs PIL image processor torchvision dependency are unrelated." + "reason": "Model-loading failure and PIL image-processor torchvision dependency are unrelated." }, { - "left": "issue:39692", - "right": "issue:44610", + "left": "issue:43531", + "right": "issue:44589", "accept": false, - "reason": "Both are multimodal, but one is docs/model-processor mismatch and quantization error while the other is an OmDet input-size mismatch." + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to missing Float8 storage support." }, { "left": "issue:43994", "right": "issue:44671", "accept": false, - "reason": "Both are accuracy/behavior complaints, but they affect different models and different code paths." + "reason": "Different models and different incorrect-output bugs; too broad to merge." }, { - "left": "issue:43531", - "right": "issue:44589", + "left": "issue:39692", + "right": "issue:44610", "accept": false, - "reason": "Sliding-window behavior and Float8 storage lookup are unrelated bugs." + "reason": "Both are vision/documentation-ish, but the concrete failures and model/processor mismatches are different." }, { "left": "issue:43421", "right": "issue:44186", "accept": false, - "reason": "Different tokenizer issues: runtime special-token updates vs LayoutLMv2 batching/NER crashes." + "reason": "Tokenizer special-token update logic is unrelated to LayoutLMv2 NER padding/truncation crashes." }, { "left": "issue:45081", "right": "issue:45356", - "accept": true, - "reason": "Same fix_mistral_regex / _patch_mistral_regex code path; both report tokenizer loading regressions around Mistral/Kimi regex patching." + "accept": false, + "reason": "Both mention `fix_mistral_regex`, but one is an AttributeError crash and the other is a codec-handling regression; not enough evidence they are the same underlying bug." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "Both mention quantization-related workflows, but one is a SigLIP2 doc/example problem and the other is offloading behavior under quantization." + "reason": "SigLIP2 doc/example issues are unrelated to quantization/offloading behavior." }, { - "left": "issue:43746", - "right": "issue:45381", + "left": "issue:43531", + "right": "issue:43645", "accept": false, - "reason": "PEFT local checkpoint loading for GraniteSpeech is unrelated to Qwen2.5-VL video position IDs." + "reason": "Sliding-window MoE behavior is unrelated to custom-model initialization in Jupyter notebooks." }, { "left": "issue:45183", "right": "issue:45479", "accept": false, - "reason": "Generic input_ids/inputs_embeds validation messaging is not the same as the single-label zero-loss classification bug." + "reason": "Input validation errors are unrelated to the zero-loss classification bug." }, { - "left": "issue:43531", - "right": "issue:43645", + "left": "issue:43746", + "right": "issue:45381", "accept": false, - "reason": "Sliding-window Qwen3-MoE and Jupyter custom-model initialization are different failures." + "reason": "PEFT checkpoint loading and Qwen2.5-VL video position IDs are unrelated." }, { "left": "issue:43596", "right": "issue:44220", "accept": false, - "reason": "Deepspeed ZeRO-3/BertModel initialization and fbank feature extraction are unrelated." + "reason": "Deepspeed/BertModel zero3 init bug is unrelated to fbank feature extraction." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "Local checkpoint loading and quantized offloading are different problems." + "reason": "PEFT checkpoint loading and quantization offload behavior are different problems." }, { "left": "issue:40444", "right": "issue:43746", "accept": false, - "reason": "Multi-image IterableDataset finetuning and GraniteSpeech checkpoint loading do not share the same underlying bug." + "reason": "Multi-image iterable dataset finetuning and GraniteSpeech PEFT checkpoint loading are unrelated." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it correctly treats the cluster as mostly unrelated, with only a narrow tokenizer-regex/offline duplication subset. The two accepted soft pairs are plausibly mergeable around the same bug family, and the rejected pairs are not overstated. No PR candidate is present, so there is nothing to merge-evaluate there." + "accept": false, + "feedback": "The summary is mostly grounded, but the soft-edge acceptance of issue:42886 \u2194 issue:44843 is too strong for the packet. The packet only gives title-level evidence, and while both mention HF_HUB_OFFLINE/tokenizer loading, there isn\u2019t enough here to confidently call them the same bug or a mergeable duplicate. A more conservative verdict would reject that pair unless the issue bodies provide stronger shared-root-cause evidence." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -4019,8 +4076,8 @@ "soft_pairs": [ "issue:43540|issue:44008", "issue:43618|issue:43697", - "issue:43716|issue:44008", "issue:43720|issue:43819", + "issue:43716|issue:44008", "issue:44220|issue:44661", "issue:43335|issue:43756", "issue:43653|issue:43720", @@ -4030,16 +4087,16 @@ "issue:43550|issue:45072", "issue:30064|issue:36010", "issue:43577|issue:43818", - "issue:43540|issue:45381", "issue:43550|issue:43927", "issue:44062|issue:44488", + "issue:43540|issue:45381", "issue:43643|issue:44877", "issue:43577|issue:43582", "issue:43531|issue:43643", "issue:43782|issue:44977", "issue:43644|issue:43749", - "issue:44448|issue:44671", "issue:43653|issue:43819", + "issue:44448|issue:44671", "issue:44561|issue:45020", "issue:43825|issue:43994", "issue:43550|issue:43819", @@ -4048,214 +4105,214 @@ "issue:43650|issue:43906", "issue:44821|issue:45084" ], - "prepared_review_unit_hash": "c89fbbdc3bdf0712ac3367b68d18b1159193e6e5a3c06a032a627d04be62533a", + "prepared_review_unit_hash": "b5b13790f087834893d5125774efa5130e85253d9da426f3f49d7c3571dc85d7", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36739, - "estimated_input_tokens": 9185, - "estimated_eval_tokens": 18626 + "serialized_chars": 36742, + "estimated_input_tokens": 9186, + "estimated_eval_tokens": 18628 }, "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 bug reports. The soft pairs share broad themes like loading, tokenizers, or model outputs, but they do not look like the same underlying bug/change, so none should be merged as duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "issue:45020 is the broadest open report and the best umbrella for the recurring remote_code regression theme among the items here.", + "summary": "This cluster is a broad mix of unrelated Transformers issues across different models, tokenizer/config, multimodal, and loading paths. The soft-similarity pairs are mostly topical overlaps, not the same underlying bug or change, so none should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the most representative issue in the set: open, broad, and aligned with several nearby remote_code-related reports.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43540", "right": "issue:44008", "accept": false, - "reason": "Different models and failures: Qwen3OmniMoe video input processing vs Gemma3n forward variable collision." + "reason": "Different failures and code paths: Qwen3OmniMoe video processing ValueError vs Gemma3n forward attribute collision." }, { "left": "issue:43618", "right": "issue:43697", "accept": false, - "reason": "Both are output regressions, but for different models and different code paths; not the same bug." + "reason": "CLIPOutput attentions missing and RTDetrV2 output drift are unrelated model behaviors." }, { - "left": "issue:43716", - "right": "issue:44008", + "left": "issue:43720", + "right": "issue:43819", "accept": false, - "reason": "Different modality and failure mode: image preprocessor/model dtype mismatch vs audio tensor attribute collision." + "reason": "Packed-weight loading/unpacking in BitNet is unrelated to DAC.from_latents vs forward STE mismatch." }, { - "left": "issue:43720", - "right": "issue:43819", + "left": "issue:43716", + "right": "issue:44008", "accept": false, - "reason": "Unrelated issues: packed-weight loading in BitNet vs DAC latent/STE mismatch." + "reason": "Both are multimodal, but one is image preprocessor/model dtype mismatch and the other is a Gemma3n variable-name bug." }, { "left": "issue:44220", "right": "issue:44661", "accept": false, - "reason": "These concern unrelated subsystems: fbank feature extraction vs tokenizer mapping/model scaffolding." + "reason": "Audio feature extraction issue vs tokenizer mapping/add-new-model failure; different subsystems." }, { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "Both involve model configuration details, but they are different architectural mistakes in different models." + "reason": "SwitchTransformers sparse-layer construction bug and Smollm3 RoPE layer dropping are distinct config/model issues." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "Tokenizer special-token decode bug and BitNet packed-weight loading bug are unrelated." + "reason": "Tokenizer special-token registration bug is unrelated to BitNet packed-weight loading." }, { "left": "issue:43421", "right": "issue:44811", "accept": false, - "reason": "Both touch tokenizers, but one is runtime post-processor updates and the other is batch_decode skip_special_tokens behavior." + "reason": "Runtime post-processor update for special tokens differs from Whisper batch_decode skip_special_tokens behavior." }, { "left": "issue:44186", "right": "issue:44898", "accept": false, - "reason": "Different tasks and models; no shared code-path or concrete bug indication." + "reason": "LayoutLMv2 tokenizer padding/NER crash is not the same as Perceiver image classification with non-default resolution." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Both mention Mistral tokenizer internals, but one is offline model_info patching and the other is a backend_tokenizer AttributeError." + "reason": "Both involve Mistral regex helpers, but one is offline model_info access and the other is a backend_tokenizer AttributeError; not the same bug." }, { "left": "issue:43550", "right": "issue:45072", "accept": false, - "reason": "Different regressions in different models: SDPA/torch.compile failure vs dtype mismatch in bfloat16 inference." + "reason": "Bamba torch.compile/SDPA failure is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Completely unrelated: image processor segmentation-map handling vs GenerationMixin import error." + "reason": "Image processor segmentation-map handling is unrelated to the GenerationMixin import error." }, { "left": "issue:43577", "right": "issue:43818", "accept": false, - "reason": "Different multimodal problems: BLIP2 dtype persistence vs Video-LLaVA temporal attention/weight sharing." - }, - { - "left": "issue:43540", - "right": "issue:45381", - "accept": false, - "reason": "Both are video-related, but they affect different Qwen model families and different bugs." + "reason": "Blip2 dtype persistence and Video-LLaVA temporal-attention/weight-sharing are different issues." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "Different model families and failure mechanisms: SDPA/compile issue vs config save/load token ID loss." + "reason": "torch.compile SDPA failure and DiaConfig save/load token-ID loss are unrelated bugs." }, { "left": "issue:44062", "right": "issue:44488", "accept": false, - "reason": "Both are tokenizer-related, but the concrete failure modes are different and not evidently the same bug." + "reason": "AddedToken special-argument conflict is unrelated to loading a specific tokenizer model." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Different Qwen multimodal failures: video-input processing error vs wrong vision_position_ids in qwen2.5-vl." }, { "left": "issue:43643", "right": "issue:44877", "accept": false, - "reason": "Both concern config loading/remote code, but one is missing fields in AutoConfig and the other is strict config blocking granite_speech." + "reason": "Missing fields with trust_remote_code and strict granite_speech config loading are distinct config-loading problems." }, { "left": "issue:43577", "right": "issue:43582", "accept": false, - "reason": "No concrete overlap beyond both being runtime errors; one is dtype handling, the other Apple Silicon warmup TypeError." + "reason": "Different dtype-related symptoms and platforms: Blip2 load dtype mismatch vs Apple Silicon allocator warmup TypeError." }, { "left": "issue:43531", "right": "issue:43643", "accept": false, - "reason": "Different issues: sliding_window behavior in Qwen3-MoE vs missing fields from trust_remote_code AutoConfig." + "reason": "Sparse-layer creation bug in SwitchTransformers is unrelated to trust_remote_code missing fields." }, { "left": "issue:43782", "right": "issue:44977", "accept": false, - "reason": "Different model/load-path failures: weight_only loading vs flash-attention generation behavior." + "reason": "Qwen3VL weight_only loading error and Qwen3.5 flash-attention generation issue are different failure modes." }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Both are loading-related, but non-persistent buffer junk vs FSDP CPU RAM-efficient loading are distinct problems." + "reason": "Non-persistent buffer junk in v5 and FSDP CPU RAM efficient loading are separate initialization/loading regressions." }, { - "left": "issue:44448", - "right": "issue:44671", + "left": "issue:43653", + "right": "issue:43819", "accept": false, - "reason": "Different model outputs in v4/v5 for Pegasus vs masked LM prediction regression in CamemBERT." + "reason": "Tokenizer special-token registration and DAC latent/forward mismatch are unrelated." }, { - "left": "issue:43653", - "right": "issue:43819", + "left": "issue:44448", + "right": "issue:44671", "accept": false, - "reason": "Tokenizer special-token registration issue and DAC latent/STE mismatch are unrelated." + "reason": "Pegasus v4/v5 output differences and CamemBERT masked-LM prediction issues affect different models and code paths." }, { "left": "issue:44561", "right": "issue:45020", "accept": false, - "reason": "Related in broad theme, but not the same concrete bug: one is a specific missing helper breaking trust_remote_code, the other is a wider remote_code regression report." + "reason": "Both mention remote_code/trust_remote_code, but one is the removed is_torch_fx_available API and the other is a broader remote_code regression report." }, { "left": "issue:43825", "right": "issue:43994", "accept": false, - "reason": "Different topics: pipeline translation error messaging vs a SigLIP2 AutoModel/pipeline result regression." + "reason": "pipeline() translation error messaging and SigLIP2 AutoModel/pipeline bad outputs are different issues." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Distinct model and failure path; no evidence they fix the same code-path." + "reason": "Bamba SDPA/torch.compile failure is unrelated to DAC.from_latents vs forward mismatch." }, { "left": "issue:44186", "right": "issue:44442", "accept": false, - "reason": "Different tokenizer/model loading bugs with no shared underlying defect." + "reason": "LayoutLMv2 tokenizer crash is not the same as AutoTokenizer failing to load FastSpeech2ConformerTokenizer." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "One is an error message bug, the other is stale docs; related theme but not the same change." + "reason": "One is an incorrect error message, the other is stale docs; related topic but not the same concrete bug/change." }, { "left": "issue:43650", "right": "issue:43906", "accept": false, - "reason": "Unrelated content; no shared bug or subsystem beyond being issue reports." + "reason": "Placeholder/nonsensical issue title versus an isolated reproduction of another bug; no shared underlying change." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "Different failures: AutoImageProcessor URL loading vs Jinja/template compilation error." + "reason": "AutoImageProcessor URL loading and template-node compilation errors are unrelated." } ] }, @@ -4332,15 +4389,15 @@ "issue:41720|issue:44991", "issue:43606|issue:45072", "issue:44805|issue:44936", - "issue:44610|issue:45072", "issue:30064|issue:36331", - "issue:42175|issue:45081", + "issue:44610|issue:45072", "issue:44485|issue:44509", + "issue:42175|issue:45081", "issue:43782|issue:45325", "issue:34634|issue:36010", - "issue:43262|issue:43819", "issue:43720|issue:45072", "issue:44060|issue:44368", + "issue:43262|issue:43819", "issue:45081|issue:45406", "issue:43749|issue:44291", "issue:43582|issue:44857", @@ -4352,27 +4409,27 @@ "issue:43454|issue:43577", "issue:34634|issue:36331", "issue:44568|issue:45488", - "issue:43746|issue:44811", "issue:43257|issue:43329", + "issue:43746|issue:44811", "issue:43638|issue:44936", "issue:43452|issue:44991", "issue:43854|issue:44977", "issue:38175|issue:44610" ], - "prepared_review_unit_hash": "1e7b9a1457423a2c4df3638377a0df392d01e81f700a37e40dd9d9cd4297d578", + "prepared_review_unit_hash": "6d93d5f0d88fbbfe741d2497c3f303007c5af952a86f8a5d4dc18d3d1a779bad", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36959, + "serialized_chars": 36957, "estimated_input_tokens": 9240, "estimated_eval_tokens": 18736 }, @@ -4381,186 +4438,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items are a heterogeneous set of unrelated bugs; none of the soft-edge pairs look like the same underlying issue or change, so all soft merges are rejected. The most triage-worthy issue is the open, well-specified Mistral tokenizer regression.", - "confidence": 0.93, - "canonical_issue_reason": "issue:45081 is the strongest anchor: it is open, has clear repro/error text, and the bug is narrowly scoped to one tokenizer patch path.", + "summary": "The soft pairs are all cross-topic regressions affecting different models, APIs, or code paths; none look like true duplicates. Most are superficially similar only at the level of broad subsystem (tokenizers, loading, dtype, processors), but the concrete failures differ.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:45081 is the best standalone issue to keep as a representative example because it is specific, reproducible, and actively referenced.", + "best_issue_reason": "issue:45081 is the strongest standalone report here: it has a clear traceback, a specific failing code path (`_patch_mistral_regex` / `backend_tokenizer`), and inbound references, making it the most actionable issue among the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41720", "right": "issue:44991", "accept": false, - "reason": "Different failures: Qwen3 device-mapping CUDA assert vs tokenizer loading for est-roberta." + "reason": "Different failures: Qwen3 auto device mapping triggers `cudaErrorAssert`, while the other is a tokenizer load failure for a different model family." }, { "left": "issue:43606", "right": "issue:45072", "accept": false, - "reason": "Bark CPU offload device mismatch and bfloat16 dtype mismatch are unrelated bugs." + "reason": "Both involve inference-time mismatches, but one is a Bark CPU-offload device mismatch and the other is a bf16 dtype mismatch in different model code paths." }, { "left": "issue:44805", "right": "issue:44936", "accept": false, - "reason": "A tensor-mask shape IndexError is not the same as train-then-evaluate failure." - }, - { - "left": "issue:44610", - "right": "issue:45072", - "accept": false, - "reason": "OmDet input-size mismatch and dtype mismatch in inference are different code paths." + "reason": "Mask shape/indexing error during model use is unrelated to `trainer.evaluate()` failing after `trainer.train()`." }, { "left": "issue:30064", "right": "issue:36331", "accept": false, - "reason": "Segmentation map processing and CustomTrainer loss signature are unrelated." + "reason": "Image processor void-segmentation handling and a custom trainer signature change are unrelated bugs." }, { - "left": "issue:42175", - "right": "issue:45081", + "left": "issue:44610", + "right": "issue:45072", "accept": false, - "reason": "Missing TensorFlow backend from install extras is unrelated to Mistral regex patching." + "reason": "Processor image-size mismatch for OmDet-Turbo is not the same bug as dtype mismatches in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:44485", "right": "issue:44509", "accept": false, - "reason": "GLM-5 RoPE discussion and removed pipeline docs are not the same bug." + "reason": "GLM-5 RoPE implementation discussion is unrelated to documentation still mentioning removed pipeline tasks." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Package extras/backend installation issue is unrelated to a tokenizer regex patch crash." }, { "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Qwen3VL loading error and Qwen2.5-VL RoPE position scaling are distinct issues." + "reason": "Different Qwen vision-language bugs: one is a load error with `weight_only=True`, the other is incorrect RoPE position scaling." }, { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "Bark voice_preset handling and GenerationMixin import failure are unrelated." - }, - { - "left": "issue:43262", - "right": "issue:43819", - "accept": false, - "reason": "Audio sampling-rate defaulting and DAC STE mismatch are different processor/model bugs." + "reason": "A BarkProcessor `voice_preset` problem is not the same as a missing `GenerationMixin` import." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "Packed-weight unpacking during accelerate loading is not a dtype mismatch problem." + "reason": "Packed-weight unpacking during accelerate loading is a different loading bug than bf16 dtype mismatches at inference." }, { "left": "issue:44060", "right": "issue:44368", "accept": false, - "reason": "These both mention tied-weights warnings, but the model/config circumstances differ." + "reason": "Both mention tied-weights warnings, but they describe different models and different root causes." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and DAC forward/from_latents mismatch are unrelated." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Missing backend_tokenizer on Tokenizer and missing _tokenizer on Gemma4Processor are different attributes and paths." + "reason": "The Mistral tokenizer regex patch crash is unrelated to Gemma4Processor missing `_tokenizer` in transformers serve." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "FSDP CPU RAM efficient loading and init_empty_weights argument handling are separate loading bugs." + "reason": "Both concern loading flows, but one is FSDP CPU RAM efficient loading and the other is an `init_empty_weights` argument regression; not the same code-path problem." }, { "left": "issue:43582", "right": "issue:44857", "accept": false, - "reason": "Apple Silicon warmup TypeError and CUDA AMP crash are unrelated runtime failures." + "reason": "Apple Silicon allocator warmup TypeError and CUDA float16 loss crash are different platform/model issues." }, { "left": "issue:43653", "right": "issue:44671", "accept": false, - "reason": "BigBird mask-token registration and CamemBERT masked LM predictions are different components." + "reason": "BigBirdTokenizer special-token registration and CamemBERT masked-LM prediction regressions are different failures." }, { "left": "issue:43526", "right": "issue:43697", "accept": false, - "reason": "BEiT label reduction and RTDetrV2 output differences are unrelated vision bugs." + "reason": "BEiT label reduction bug and RTDetrV2 output changes in v5 are unrelated vision issues." }, { "left": "issue:43526", "right": "issue:44554", "accept": false, - "reason": "BEiT label handling and MPS attention correctness are not the same issue." + "reason": "Image processor label reduction bug is unrelated to the MPS attention correctness issue." }, { "left": "issue:43653", "right": "issue:44912", "accept": false, - "reason": "Tokenizer special-token registration and MXFP4 quantization fallback are unrelated." + "reason": "Tokenizer special-token decoding bug and MXFP4 quantization fallback are unrelated." }, { "left": "issue:45084", "right": "issue:45538", "accept": false, - "reason": "Template compilation TypeError and CLIPTokenizer max-length behavior are different failures." + "reason": "Template compilation error and CLIPTokenizer `model_max_length` are unrelated tokenizer/config issues." }, { "left": "issue:43454", "right": "issue:43577", "accept": false, - "reason": "AyaVision tied-weight loading and Blip2 dtype propagation are separate model-loading bugs." + "reason": "AyaVision tied-weight loading and Blip2 dtype persistence are different model-loading bugs." }, { "left": "issue:34634", "right": "issue:36331", "accept": false, - "reason": "BarkProcessor preset support and trainer compute_loss kwargs have no overlap." + "reason": "Repeated pair: BarkProcessor preset issue and trainer loss-signature issue are unrelated." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "MDeBERTa add_special_tokens regression and Llama tokenizer pre-tokenizer override are distinct tokenizer issues." + "reason": "Both are tokenizer regressions, but one is missing BOS/EOS addition and the other overrides ByteLevel pre-tokenizer behavior; different concrete defects." }, { - "left": "issue:43746", - "right": "issue:44811", + "left": "issue:43257", + "right": "issue:43329", "accept": false, - "reason": "PEFT checkpoint loading and Whisper batch_decode skip_special_tokens behavior are unrelated." + "reason": "MOE weight conversion during accelerate+deepspeed loading is unrelated to the undefined video-token counting helpers." }, { - "left": "issue:43257", - "right": "issue:43329", + "left": "issue:43746", + "right": "issue:44811", "accept": false, - "reason": "Qwen3 MOE conversion under accelerate/deepspeed and multimodal token counting are different bugs." + "reason": "PEFT local checkpoint loading and Whisper `batch_decode()` special-token handling are different bugs." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "Zero3/Bert initialization IndexError and post-train evaluate failure are not the same problem." + "reason": "Deepspeed ZeRO-3 empty tensor indexing and `trainer.evaluate()` breaking after training are unrelated training-flow issues." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "gguf_file handling and tokenizer loading for a specific RoBERTa model are different loading regressions." + "reason": "GGUF loading support and tokenizer loading for `EMBEDDIA/est-roberta` are separate `from_pretrained` failures." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "GLM-4.7-Flash test loading and Qwen3.5 flash-attention generation differ in model and symptom." + "reason": "GLM-4.7-Flash unit-test loading and Qwen3.5 flash-attention generation problems are different model-specific bugs." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Siglip2 zero probabilities and OmDet input resolution mismatch are unrelated." + "reason": "Both are vision-model regressions, but one is zero probabilities in SigLIP2 and the other is a processor/model input-size mismatch." } ] }, @@ -4611,7 +4668,6 @@ "issue:43749", "issue:43756", "issue:43819", - "issue:43906", "issue:43927", "issue:43931", "issue:44291", @@ -4627,17 +4683,19 @@ "issue:44987", "issue:45042", "issue:45072", + "issue:45137", "issue:45356", "issue:45357", "issue:45381", + "issue:45405", "issue:45538" ], "soft_pairs": [ "issue:44589|issue:45042", "issue:39692|issue:44912", "issue:43257|issue:43696", - "issue:43262|issue:44811", "issue:43819|issue:43927", + "issue:43262|issue:44811", "issue:43720|issue:44610", "issue:43550|issue:44912", "issue:36010|issue:43645", @@ -4646,9 +4704,9 @@ "issue:43329|issue:44479", "issue:43531|issue:45042", "issue:43257|issue:43931", - "issue:43329|issue:43450", "issue:36010|issue:36331", "issue:43257|issue:43756", + "issue:43329|issue:43450", "issue:43577|issue:45072", "issue:43697|issue:45357", "issue:42175|issue:44938", @@ -4657,235 +4715,231 @@ "issue:44479|issue:45381", "issue:43927|issue:44912", "issue:33357|issue:34567", - "issue:44987|issue:45538", "issue:42915|issue:43742", + "issue:44987|issue:45538", "issue:44291|issue:45042", - "issue:43540|issue:44560", "issue:43317|issue:45538", - "issue:43540|issue:44479", - "issue:43906|issue:45381" + "issue:43540|issue:44560", + "issue:45137|issue:45405" ], - "prepared_review_unit_hash": "539a559b3dc4bec60e66ab6156045592b20918468a69e82cf25c5fb7217480a6", + "prepared_review_unit_hash": "487e66e3dbfa979f457d3e0c48ed4a6d0ad690e2d815355e593a3609affce0ad", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 36534, - "estimated_input_tokens": 9134, - "estimated_eval_tokens": 18524 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36778, + "estimated_input_tokens": 9195, + "estimated_eval_tokens": 18646 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the issue titles span unrelated bugs in tokenizers, multimodal/video handling, loading/quantization, packaging, and trainer APIs. I don\u2019t see a single duplicate family that safely unifies these items, so all soft-edge candidates are rejected.", - "confidence": 0.28, - "canonical_issue_reason": "No clear canonical duplicate exists across the cluster because the issues cover unrelated subsystems and failure modes. If forced to pick a representative issue, issue 44479 is the broadest and most centrally described regression report, but it is not a true cluster center.", + "summary": "This cluster is mostly a grab bag of unrelated issues across loading, tokenizer, multimodal/video, training, dtype, and packaging bugs. There is one close Qwen video-regression subfamily, but the rest are not duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 44479 is the broadest representative of the only clearly related subgroup (Qwen video-input regressions), but the overall cluster is not a true duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44479 is the best single representative only because it has the broadest impact statement and a concrete regression framing across multiple Qwen video models. It is still not a good canonical for the full set, since the rest are unrelated.", + "best_issue_reason": "Issue 44479 is the best representative issue because it captures the widest blast radius among the video-regression reports, though it still does not subsume the rest of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44589", "right": "issue:45042", "accept": false, - "reason": "Different bugs: float8 storage loading vs PIL backend torchvision dependency." + "reason": "Different bug classes: float8 storage/load failure vs PIL backend dependency checking." }, { "left": "issue:39692", "right": "issue:44912", "accept": false, - "reason": "Unrelated topics: documentation/example errors vs MXFP4 quantization fallback." + "reason": "Doc-example/model-processor mismatch and quantization failure are unrelated to MXFP4 loading fallback." }, { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Different model-loading/Deepspeed conversion bug vs plain CUDA OOM." + "reason": "Qwen3 MoE weight-conversion bug and GPT-oss OOM are different problems." }, { - "left": "issue:43262", - "right": "issue:44811", + "left": "issue:43819", + "right": "issue:43927", "accept": false, - "reason": "Both are audio-related, but one is chat-template sampling rate handling and the other is batch_decode skip_special_tokens behavior." + "reason": "DAC latent reconstruction and DiaConfig token-ID persistence are unrelated code paths." }, { - "left": "issue:43819", - "right": "issue:43927", + "left": "issue:43262", + "right": "issue:44811", "accept": false, - "reason": "Different code paths: DAC latent reconstruction vs DiaConfig save/load token ID persistence." + "reason": "Audio chat-template sampling-rate default and Whisper batch_decode skip_special_tokens are different processor bugs." }, { "left": "issue:43720", "right": "issue:44610", "accept": false, - "reason": "Different subsystems: BitNet accelerate loading vs OmDet-Turbo image-size mismatch." + "reason": "Packed-weight unloading during accelerate load is unrelated to an OmDet-Turbo image-size mismatch." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "Different failures: torch.compile/SDPA on Bamba vs quantization fallback on git-oss." + "reason": "torch.compile/SDPA failure and MXFP4 quantization loading are separate issues." }, { "left": "issue:36010", "right": "issue:43645", "accept": false, - "reason": "Both involve Transformers v5 breakage, but one is a GenerationMixin import error and the other is custom model initialization in notebooks." + "reason": "GenerationMixin import breakage is not the same as Jupyter custom-model initialization." }, { "left": "issue:44568", "right": "issue:45356", "accept": false, - "reason": "Different tokenizer regressions affecting different models and symptoms." + "reason": "Tokenizer special-token insertion and codec-handling/warning regression are distinct tokenizer bugs." }, { "left": "issue:43749", "right": "issue:44756", "accept": false, - "reason": "Both are memory/loading issues, but one is FSDP CPU RAM efficient loading and the other is mmap on Strix Halo." + "reason": "FSDP CPU RAM efficient loading and Strix Halo mmap OOM are different memory-loading problems." }, { "left": "issue:43329", "right": "issue:44479", "accept": false, - "reason": "Related video area, but 43329 is an undefined-function bug in token counting while 44479 is a broader video-input regression across models." + "reason": "Undefined video-helper references are a different failure mode from the broader Qwen video-input regression." }, { "left": "issue:43531", "right": "issue:45042", "accept": false, - "reason": "Unrelated: Qwen3-MoE sliding window vs PIL backend torchvision dependency." + "reason": "Qwen3-MoE sliding-window behavior and PIL/torchvision dependency handling are unrelated." }, { "left": "issue:43257", "right": "issue:43931", "accept": false, - "reason": "Different model-loading issues: Qwen3 MoE weight conversion vs Qwen3-VL weight shape mismatch." - }, - { - "left": "issue:43329", - "right": "issue:43450", - "accept": false, - "reason": "Both touch video processing, but one is undefined helpers in token counting and the other is batched video shape output." + "reason": "MoE weight conversion and Qwen3-VL weight-shape mismatch are not the same bug." }, { "left": "issue:36010", "right": "issue:36331", "accept": false, - "reason": "Different API breakages: import path change vs Trainer.compute_loss signature change." + "reason": "Importing GenerationMixin and CustomTrainer.compute_loss signature changes are unrelated." }, { "left": "issue:43257", "right": "issue:43756", "accept": false, - "reason": "Completely different bugs: Qwen3 MoE loading vs Smollm3 RoPE-layer count." + "reason": "Qwen3 MoE accelerate loading and Smollm3 RoPE-layer dropping are different model issues." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Undefined multimodal-token helpers and batched video-shape errors are separate video-processing bugs." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both mention dtype, but they affect different models and different failure modes." + "reason": "Blip2 dtype sticking in float32 and SwitchTransformers/TimmWrapper dtype mismatches are distinct." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "Different regressions: RTDetrV2 inference drift vs Qwen3.5 save_pretrained key corruption." + "reason": "RTDetrV2 inference divergence and Qwen3.5 save_pretrained key mapping are unrelated." }, { "left": "issue:42175", "right": "issue:44938", "accept": false, - "reason": "Unrelated packaging/runtime issues: missing tensorflow backend vs Python 3.14 import/load failure." + "reason": "Packaging missing TensorFlow and Python 3.14 import/load failure are different issues." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "Different areas: gguf tokenizer/model loading vs Apple Silicon allocator warmup TypeError." + "reason": "gguf_file loading breakage and Apple Silicon warmup TypeError are unrelated." }, { "left": "issue:41720", "right": "issue:43540", "accept": false, - "reason": "Different Qwen-related failures: auto device mapping CUDA assert vs video input ValueError." + "reason": "Auto device-mapping CUDA assert and Qwen3OmniMoe video-processing ValueError are different code paths." }, { "left": "issue:44479", "right": "issue:45381", - "accept": false, - "reason": "Both are video regressions in Qwen models, but one is a broad multi-model regression and the other is a specific vision_position_ids bug; not enough to treat as the same issue." + "accept": true, + "reason": "These describe the same Qwen video-input regression family: both center on incorrect video position-id handling in 5.3.0." }, { "left": "issue:43927", "right": "issue:44912", "accept": false, - "reason": "Different bugs: DiaConfig token-ID persistence vs MXFP4 quantization loading." + "reason": "DiaConfig custom-token persistence and MXFP4 loading fallback are unrelated." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "Different failures in unrelated areas: Mac bus error on CLIP model vs TrainerState token counter update." - }, - { - "left": "issue:44987", - "right": "issue:45538", - "accept": false, - "reason": "Different tokenizer/load bugs with no shared code-path evidence." + "reason": "MacOS bus error with CLIP and TrainerState token counting are completely unrelated." }, { "left": "issue:42915", "right": "issue:43742", "accept": false, - "reason": "Different model-loading failures: Qwen3Moe FP8 config vs MobileLLM key error." + "reason": "Qwen3Moe FP8 config failure and MobileLLM key error are different loading bugs." }, { - "left": "issue:44291", - "right": "issue:45042", + "left": "issue:44987", + "right": "issue:45538", "accept": false, - "reason": "Unrelated issues: init_empty_weights argument handling vs PIL backend torchvision dependency." + "reason": "Model loading regression for physical-intelligence/fast and CLIPTokenizer max_length bug are unrelated." }, { - "left": "issue:43540", - "right": "issue:44560", + "left": "issue:44291", + "right": "issue:45042", "accept": false, - "reason": "Both involve Qwen video input, but the reported errors and model families differ; not the same concrete bug." + "reason": "init_empty_weights argument compatibility and PIL backend torchvision requirement are different bugs." }, { "left": "issue:43317", "right": "issue:45538", "accept": false, - "reason": "Different categories: dequantized model offload failure vs CLIPTokenizer model_max_length behavior." + "reason": "Device-map/offload dequantization failure and CLIPTokenizer model_max_length are unrelated." }, { "left": "issue:43540", - "right": "issue:44479", + "right": "issue:44560", "accept": false, - "reason": "Related surface area, but the exact failures differ and there is no evidence they are the same code-path bug." + "reason": "Both are video-related, but Qwen3OmniMoe ValueError and Qwen3-vl-embedding StopIteration are different model/code-path failures." }, { - "left": "issue:43906", - "right": "issue:45381", + "left": "issue:45137", + "right": "issue:45405", "accept": false, - "reason": "Unrelated: isolated reproduction of issue 38071 vs Qwen2.5-VL video position-id regression." + "reason": "DeepSpeed ZeRO3 deque error and PEFT minimum-version bump are unrelated." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative: it treats the cluster as mostly unrelated while recognizing the one plausible Qwen video-regression subgroup. The accepted soft edge (44479/45381) is supported by the titles, and the rejected pairs do not overclaim duplication based on the available evidence." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -4914,6 +4968,7 @@ "issue:43452", "issue:43493", "issue:43526", + "issue:43540", "issue:43577", "issue:43582", "issue:43643", @@ -4925,9 +4980,11 @@ "issue:43756", "issue:43782", "issue:43824", + "issue:43906", "issue:43994", "issue:44062", "issue:44291", + "issue:44479", "issue:44484", "issue:44488", "issue:44514", @@ -4939,34 +4996,32 @@ "issue:44898", "issue:44964", "issue:44991", - "issue:45042", "issue:45081", - "issue:45137", "issue:45216", "issue:45290", "issue:45325", "issue:45375", "issue:45381", - "issue:45405", "issue:45479", "issue:45538" ], "soft_pairs": [ - "issue:45137|issue:45405", + "issue:43906|issue:45381", + "issue:43540|issue:44479", "issue:42915|issue:45538", - "issue:43526|issue:45479", - "issue:42175|issue:43824", "issue:42915|issue:44821", + "issue:42175|issue:43824", + "issue:43526|issue:45479", "issue:42175|issue:44488", "issue:44062|issue:44991", "issue:38175|issue:44898", "issue:28282|issue:34689", "issue:43646|issue:43749", "issue:43582|issue:43782", - "issue:43493|issue:43994", "issue:28282|issue:29127", - "issue:43643|issue:45375", + "issue:43493|issue:43994", "issue:43742|issue:44291", + "issue:43643|issue:45375", "issue:43756|issue:45325", "issue:36010|issue:42175", "issue:44484|issue:44704", @@ -4977,239 +5032,232 @@ "issue:44062|issue:45081", "issue:43688|issue:43749", "issue:44514|issue:45290", - "issue:43452|issue:43577", "issue:43643|issue:44561", - "issue:44514|issue:45381", + "issue:43452|issue:43577", "issue:43742|issue:44589", - "issue:43697|issue:45216", - "issue:45042|issue:45081" + "issue:44514|issue:45381", + "issue:43697|issue:45216" ], - "prepared_review_unit_hash": "013167a260c9932fe10db7a9308351e78b59431363969f35639596580fbf939b", + "prepared_review_unit_hash": "59ad27e1c4710c10d6515dc43c55c21d6112e7f0adce329efbe367e6a05ea629", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37189, - "estimated_input_tokens": 9298, - "estimated_eval_tokens": 18852 + "serialized_chars": 37231, + "estimated_input_tokens": 9308, + "estimated_eval_tokens": 18872 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of Transformers issues. Only two soft duplicate pockets look plausible: SigLIP2 bad outputs/discrepancy reports, and Qwen2.5-VL vision position-id / rope-index bugs. The rest are clearly different bugs or features.", - "confidence": 0.66, - "canonical_issue_reason": "Issue 43493 is the clearest root-cause-style report in the strongest duplicate pocket, explicitly comparing HF vs original JAX behavior for SigLIP2.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers regressions and model-specific loading bugs. Only a couple of pairs look like they share the same underlying code path; most should stay separate.", + "confidence": 0.39, + "canonical_issue_reason": "Issue 43646 is the broadest umbrella item here: a v5.0 custom-model initialization regression. It best matches the cluster\u2019s recurring theme of version-related loading/init breakages, even though many other issues are too specific to be true duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "43493 is the best overall triage target because it describes the underlying implementation mismatch directly, rather than only a downstream symptom.", + "best_issue_reason": "43646 is the most representative issue for this cluster because it is the most general and central failure mode among the mostly version-regression/loading reports. None of the other issues is as good an umbrella candidate.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45137", - "right": "issue:45405", + "left": "issue:43906", + "right": "issue:45381", + "accept": false, + "reason": "Both mention Qwen2.5-VL-style regressions, but one is an isolated reproduction of a different issue and the other is a specific v5.3.0 position-id bug. Not the same underlying problem." + }, + { + "left": "issue:43540", + "right": "issue:44479", "accept": false, - "reason": "DeepSpeed ZeRO3 deque error vs PEFT version bump; unrelated problems." + "reason": "Both are video-input problems in Qwen-family models, but they involve different model families and different symptoms. Too broad to merge." }, { "left": "issue:42915", "right": "issue:45538", "accept": false, - "reason": "FP8 config failure and CLIP tokenizer max length are different code paths." + "reason": "Qwen3Moe FP8 failure and CLIPTokenizer max_length are unrelated bugs." }, { - "left": "issue:43526", - "right": "issue:45479", + "left": "issue:42915", + "right": "issue:44821", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug is unrelated to sequence-classification loss degeneracy." + "reason": "Different subsystems and different failures; no shared underlying bug." }, { "left": "issue:42175", "right": "issue:43824", "accept": false, - "reason": "Missing TensorFlow backend in a torch install is unrelated to a missing model class import." + "reason": "One is backend/package installation, the other is a model import failure. Not the same issue." }, { - "left": "issue:42915", - "right": "issue:44821", + "left": "issue:43526", + "right": "issue:45479", "accept": false, - "reason": "Different models and failures; not the same underlying bug." + "reason": "BEiT label reduction and zero-loss classification are different code paths and model families." }, { "left": "issue:42175", "right": "issue:44488", "accept": false, - "reason": "Backend packaging issue vs a model-loading failure for a specific model." + "reason": "Packaging/backend availability versus loading a specific model; unrelated causes." }, { "left": "issue:44062", "right": "issue:44991", "accept": false, - "reason": "AddedToken keyword collision and tokenizer loading failure are not clearly the same concrete bug." + "reason": "Different tokenizer/loading failures with no clear shared root cause." }, { "left": "issue:38175", "right": "issue:44898", "accept": false, - "reason": "Different model families and symptoms; too broad to call duplicates." + "reason": "SigLIP2 zero-probabilities and Perceiver interpolation failure are different models and bugs." }, { "left": "issue:28282", "right": "issue:34689", "accept": false, - "reason": "PyTorch-missing ImportError vs Llama 3.2 Vision loading regression are different issues." + "reason": "Missing PyTorch dependency and Llama 3.2 model loading regression are not the same problem." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Custom model init breakage and FSDP CPU RAM loading are separate failures." + "reason": "Generic custom-model init breakage and FSDP CPU RAM-efficient loading are different loading paths." }, { "left": "issue:43582", "right": "issue:43782", "accept": false, - "reason": "Apple Silicon allocator TypeError vs Qwen3VL weight_only load error are different code paths." + "reason": "Apple Silicon allocator TypeError and Qwen3VL weight_only loading error are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Environment dependency error versus a LayoutLMv3 error-message-quality issue." }, { "left": "issue:43493", "right": "issue:43994", "accept": true, - "reason": "Both are SigLIP2 base-patch16-224 output correctness regressions and likely stem from the same model implementation bug." + "reason": "Both target google/siglip2-base-patch16-224 and describe incorrect outputs from the HF SigLIP2 implementation; the JAX discrepancy and nonsensical pipeline results look like the same underlying model bug." }, { - "left": "issue:28282", - "right": "issue:29127", + "left": "issue:43742", + "right": "issue:44291", "accept": false, - "reason": "Missing PyTorch ImportError is unrelated to LayoutLMv3 box-validation messaging." + "reason": "Different load-time TypeErrors with different triggers and code paths." }, { "left": "issue:43643", "right": "issue:45375", "accept": false, - "reason": "Missing fields with trust_remote_code and a dropped config field are similar only at a high level, not the same bug." - }, - { - "left": "issue:43742", - "right": "issue:44291", - "accept": false, - "reason": "Model key lookup failure and init_empty_weights argument mismatch are different loading failures." + "reason": "Both involve config-field handling, but one is a generic trust_remote_code config-loss issue and the other is a Qwen3_5MoeVisionConfig field omission. Too different." }, { "left": "issue:43756", "right": "issue:45325", "accept": false, - "reason": "Different Qwen2.5-VL bugs: RoPE-layer omission vs position-id scaling." + "reason": "Both are Qwen-related, but one is about SmolLM3 RoPE layers and the other is a Qwen2.5-VL position-id regression." }, { "left": "issue:36010", "right": "issue:42175", "accept": false, - "reason": "GenerationMixin import error and TensorFlow backend packaging are unrelated." + "reason": "GenerationMixin import failure and TensorFlow backend packaging are unrelated." }, { "left": "issue:44484", "right": "issue:44704", "accept": false, - "reason": "save_pretrained shard-size behavior and AutoProcessor kwargs forwarding are not the same issue." + "reason": "save_pretrained shard-size semantics and AutoProcessor cached_file kwargs are different API bugs." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "TrainerState token counter update and CustomTrainer loss signature breakage are distinct trainer issues." + "reason": "TrainerState token counting and custom Trainer compute_loss signature are separate trainer issues." }, { "left": "issue:44291", "right": "issue:44821", "accept": false, - "reason": "Init-empty-weights error and AutoImageProcessor URL loading failure are different problems." + "reason": "init_empty_weights argument handling and AutoImageProcessor URL loading are unrelated." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "Different models and failing operations; no evidence of a shared root cause." + "reason": "Janus test failure and Phi-4 multimodal load failure are both multimodal, but not the same concrete bug." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "gguf loading breakage and a specific model load failure are not the same bug." + "reason": "gguf_file handling and loading cjvt/sleng-bert are different model-loading failures." }, { "left": "issue:44062", "right": "issue:45081", "accept": false, - "reason": "Tokenizer AddedToken duplication and Mistral regex patch crash are unrelated." + "reason": "AddedToken keyword conflict and Mistral regex patch crash are different tokenizer bugs." }, { "left": "issue:43688", "right": "issue:43749", "accept": false, - "reason": "Auxiliary-loss normalization and FSDP CPU RAM efficient loading are separate concerns." + "reason": "Auxiliary-loss normalization and FSDP CPU RAM-efficient loading are unrelated." }, { "left": "issue:44514", "right": "issue:45290", "accept": false, - "reason": "Both involve apply_chat_template, but the crashes are in different scenarios and code paths." - }, - { - "left": "issue:43452", - "right": "issue:43577", - "accept": false, - "reason": "gguf loading failure and Blip2 dtype propagation are unrelated." + "reason": "Both are apply_chat_template crashes, but the triggers differ: batched padding=False versus assistant tool-call messages with no content." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Both mention trust_remote_code, but one is missing fields and the other is a v5 import removal breakage." + "reason": "Both involve trust_remote_code/custom-model behavior, but one is missing config fields and the other is removal of is_torch_fx_available. Different root causes." }, { - "left": "issue:44514", - "right": "issue:45381", + "left": "issue:43452", + "right": "issue:43577", "accept": false, - "reason": "apply_chat_template batching bug and Qwen2.5-VL vision-position-id bug are different features." + "reason": "Different loading bugs: gguf_file handling versus BLIP2 dtype propagation." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "Model key error and float8 storage lookup error are not the same issue." + "reason": "A key error when loading MobileLLM and a missing Float8 storage type are different failures." }, { - "left": "issue:43697", - "right": "issue:45216", + "left": "issue:44514", + "right": "issue:45381", "accept": false, - "reason": "RTDetrV2 output regression and Qwen3.5 save_pretrained regression are unrelated." + "reason": "Both mention Qwen2.5-VL chat/image/video behavior, but one is chat-template batching and the other is a vision-position-id regression. Not the same bug." }, { - "left": "issue:45042", - "right": "issue:45081", + "left": "issue:43697", + "right": "issue:45216", "accept": false, - "reason": "Torchvision requirement regression and Mistral regex patch crash are unrelated." - }, - { - "left": "issue:45325", - "right": "issue:45381", - "accept": true, - "reason": "Both report Qwen2.5-VL position-id / RoPE-index errors in transformers 5.3.0 and likely share the same code path." + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 save_pretrained corruption are separate regressions." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative about the two plausible duplicate pockets. The accepted soft pairs are justified by closely matching model/symptom titles, and the rejected pairs mostly differ by model, feature, or failure mode. No obvious overstatement of evidence." + "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The single accepted pair (issue:43493 / issue:43994) is at least plausibly the same SigLIP2 bug, and the rejections avoid overclaiming duplicate status for unrelated model- or subsystem-specific failures. The choice of issue:43646 as the umbrella issue is also defensible given the cluster\u2019s strong loading/init regression theme." }, "evaluator_used": true, "retried": false, @@ -5244,7 +5292,6 @@ "issue:41762", "issue:43122", "issue:43295", - "issue:43329", "issue:43452", "issue:43525", "issue:43531", @@ -5262,7 +5309,6 @@ "issue:44464", "issue:44466", "issue:44479", - "issue:44554", "issue:44568", "issue:44625", "issue:44743", @@ -5275,11 +5321,13 @@ "issue:44977", "issue:44991", "issue:45005", - "issue:45325", + "issue:45042", + "issue:45081", "issue:45356", "issue:45440" ], "soft_pairs": [ + "issue:45042|issue:45081", "issue:43531|issue:44821", "issue:44625|issue:44877", "issue:43819|issue:44811", @@ -5289,11 +5337,11 @@ "issue:43452|issue:44843", "issue:44466|issue:45005", "issue:44464|issue:44743", - "issue:41762|issue:44936", "issue:43295|issue:45356", + "issue:41762|issue:44936", "issue:24643|issue:29127", - "issue:40444|issue:44811", "issue:41720|issue:43906", + "issue:40444|issue:44811", "issue:43525|issue:43643", "issue:44361|issue:44442", "issue:38175|issue:43295", @@ -5308,224 +5356,216 @@ "issue:44977|issue:45356", "issue:30064|issue:33666", "issue:44464|issue:44977", - "issue:43653|issue:43927", - "issue:43329|issue:45325", - "issue:43577|issue:44554" + "issue:43653|issue:43927" ], - "prepared_review_unit_hash": "366c895882253a6f0335f236fa9a698010028a59c61e43b3bb09a110cb247963", + "prepared_review_unit_hash": "8b660bb64fd7ed59c709b4f732b0fc477f9b51faf553aa3468fabdfddd4e6064", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37378, - "estimated_input_tokens": 9345, - "estimated_eval_tokens": 18946 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36469, + "estimated_input_tokens": 9118, + "estimated_eval_tokens": 18492 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster does not form a duplicate set: the soft pairs span unrelated subsystems (tokenizers, vision/video processors, training/DeepSpeed, model configs, and generation regressions). None of the candidate edges looks like the same underlying bug/change.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "These items are mostly unrelated issues across different models and subsystems. None of the soft edges are strong enough to merge as true duplicates; at most they share broad themes like tokenization, config loading, or generation regressions.", + "confidence": 0.62, + "canonical_issue_reason": "No single canonical issue clearly covers the set; if forced, issue:44843 is the closest hub because it describes a concrete tokenizer-loading regression in a narrow code path that is easy to triage.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:44843 is the strongest standalone issue: it has a specific failure mode, a narrow affected path (_patch_mistral_regex), and a clear regression trigger (HF_HUB_OFFLINE).", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Different subsystems: PIL/image-processor torchvision dependency vs a Mistral tokenizer regex crash." + }, { "left": "issue:43531", "right": "issue:44821", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior is unrelated to AutoImageProcessor URL loading." + "reason": "Both involve loading behavior, but one is Qwen3-MoE sliding-window logic and the other is AutoImageProcessor URL loading." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Qwen3.5 num_labels propagation is a different config bug than granite_speech strict-config loading." + "reason": "Both are config-related, but one is num_labels propagation and the other is strict config loading for granite_speech." }, { "left": "issue:43819", "right": "issue:44811", "accept": false, - "reason": "DAC latent conversion vs Whisper batch_decode are unrelated code paths and symptoms." + "reason": "Unrelated code paths: DAC latent conversion vs Whisper processor batch_decode behavior." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "BitNet packed-weight loading is unrelated to Perceiver image classification at non-default resolution." + "reason": "Different models and bugs: packed-weight loading for BitNet vs Perceiver image-classification resolution handling." }, { "left": "issue:43688", "right": "issue:45440", "accept": false, - "reason": "Aux-loss normalization in MoE models is unrelated to DeepseekV3MoE implementation divergence." + "reason": "Both are MoE-related, but they concern different models and different failure modes, not the same bug." }, { "left": "issue:43653", "right": "issue:44568", "accept": false, - "reason": "BigBirdTokenizer special-token registration is unrelated to mdeberta-v3 BOS/EOS handling." + "reason": "Both touch special tokens, but one is BigBird mask-token registration and the other is BOS/EOS insertion for a different tokenizer." }, { "left": "issue:43452", "right": "issue:44843", "accept": false, - "reason": "gguf_file loading failures and offline model_info calls are different tokenizer/model-loading regressions." + "reason": "Both affect tokenizer/model loading, but one is gguf_file handling and the other is an offline Mistral regex patch regression." }, { "left": "issue:44466", "right": "issue:45005", "accept": false, - "reason": "Serialized lm_head weight inconsistency is not the same issue as tied-weights problems on translation models." + "reason": "Both mention tied weights, but one is device-dependent serialization and the other is translation-model behavior in v5." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Compiled-forward chunked generation mismatch is unrelated to Qwen3.5 recurrent-state reset logic." + "reason": "Both are generation regressions, but compiled-forward chunking and recurrent-state reset are different concrete bugs." }, { - "left": "issue:41762", - "right": "issue:44936", + "left": "issue:43295", + "right": "issue:45356", "accept": false, - "reason": "Gemma3 ZeRO-3 loading failure is unrelated to trainer.evaluate() breaking after train()." + "reason": "Both are regressions, but one is processor/tokenizer access in custom code and the other is Kimi-K2.5 codec handling." }, { - "left": "issue:43295", - "right": "issue:45356", + "left": "issue:41762", + "right": "issue:44936", "accept": false, - "reason": "processor.tokenizer regression with images is unrelated to Kimi-K2.5 tokenizer codec/regression handling." + "reason": "Different failures: DeepSpeed ZeRO-3 load-time IndexError vs trainer.evaluate() failing after training." }, { "left": "issue:24643", "right": "issue:29127", "accept": false, - "reason": "DeepSpeed weight-dimension runtime error and LayoutLMv3 box-validation messaging are distinct problems." + "reason": "DeepSpeed training RuntimeError and LayoutLMv3 box-validation messaging are unrelated." }, { - "left": "issue:40444", - "right": "issue:44811", + "left": "issue:41720", + "right": "issue:43906", "accept": false, - "reason": "IterableDataset multi-image finetuning failure is unrelated to Whisper batch_decode skip_special_tokens behavior." + "reason": "Auto device mapping cuda assert and an isolated reproduction of another issue do not match the same underlying bug." }, { - "left": "issue:41720", - "right": "issue:43906", + "left": "issue:40444", + "right": "issue:44811", "accept": false, - "reason": "Qwen3 auto device mapping cudaErrorAssert is unrelated to the isolated reproduction referenced by issue 38071." + "reason": "Different areas: multi-image IterableDataset finetuning vs Whisper batch_decode special-token handling." }, { "left": "issue:43525", "right": "issue:43643", "accept": false, - "reason": "Llama4Config missing pad_token_id and AutoConfig missing fields with trust_remote_code are different config-loading bugs." + "reason": "Related only at a high level; one is a specific Llama4Config pad_token_id failure, the other is generic trust_remote_code config hydration." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "MLukeTokenizer task AttributeError is unrelated to AutoTokenizer failing on FastSpeech2ConformerTokenizer." + "reason": "Different tokenizer failures with different affected components and error surfaces." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "SigLIP2 zero probabilities are a model inference issue, not the processor.tokenizer regression." + "reason": "Unexpected zero probabilities in SigLIP2 is not the same bug as the processor/tokenizer regression in v4.57.5." }, { "left": "issue:43577", "right": "issue:43697", "accept": false, - "reason": "Blip2 dtype propagation and RTDetrV2 output drift are unrelated model-specific bugs." + "reason": "BLIP-2 dtype propagation and RTDetrV2 output drift are separate model correctness issues." }, { "left": "issue:43653", "right": "issue:44361", "accept": false, - "reason": "BigBirdTokenizer special-token decode issue is unrelated to MLukeTokenizer task failures." + "reason": "Tokenizer special-token registration and MLukeTokenizer AttributeError are unrelated." }, { "left": "issue:44843", "right": "issue:44991", "accept": false, - "reason": "HF_HUB_OFFLINE model_info call in Mistral regex patch is unrelated to EMBEDDIA/est-roberta tokenizer loading." + "reason": "Both are tokenizer-load regressions, but one is offline Mistral regex patching and the other is est-roberta loading on v5." }, { "left": "issue:28282", "right": "issue:32090", "accept": false, - "reason": "PyTorch-not-found import error is unrelated to NoneType broadcast failures in Trainer." + "reason": "Missing-PyTorch import error and Trainer GPU broadcast TypeError are different problems." }, { "left": "issue:28282", "right": "issue:35141", "accept": false, - "reason": "PyTorch import absence is unrelated to tied-weight reinitialization after resizing embeddings." + "reason": "Import-time missing dependency vs tied-weight reinitialization are unrelated." }, { "left": "issue:28282", "right": "issue:33357", "accept": false, - "reason": "Missing PyTorch import error is unrelated to MacOS bus error with a community CLIP model." + "reason": "Missing-PyTorch ImportError does not match the MacOS bus-error CLIP regression." }, { "left": "issue:44479", "right": "issue:44936", "accept": false, - "reason": "Video-input regression in v5.3.0 is unrelated to trainer.evaluate() failing after training." + "reason": "Qwen video-input regression and trainer.evaluate() failure are different issues." }, { "left": "issue:43122", "right": "issue:43906", "accept": false, - "reason": "Tokenization differences across versions are unrelated to the isolated reproduction of issue 38071." + "reason": "Both mention tokenizer behavior, but one is a version-to-version tokenization change and the other is only an isolated reproduction of a different issue." }, { "left": "issue:44977", "right": "issue:45356", "accept": false, - "reason": "Qwen3.5 flash-attention generation failure is unrelated to Kimi-K2.5 tokenizer regression." + "reason": "Flash-attention generation failure and tokenizer codec regression are separate problems." }, { "left": "issue:30064", "right": "issue:33666", "accept": false, - "reason": "Void segmentation map processing and Qwen2-VL multi-GPU training are different vision/model-training issues." + "reason": "Void segmentation-map processing and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:44464", "right": "issue:44977", "accept": false, - "reason": "Chunked generation with compiled forward is unrelated to Qwen3.5 flash-attention generation failures." + "reason": "Compiled-forward chunked generation inconsistency is not the same as the flash-attention generation bug." }, { "left": "issue:43653", "right": "issue:43927", "accept": false, - "reason": "BigBirdTokenizer special-token decode bug is unrelated to DiaConfig losing custom token IDs on save/load." - }, - { - "left": "issue:43329", - "right": "issue:45325", - "accept": false, - "reason": "Undefined video-branch helpers in multimodal token counting are unrelated to Qwen2.5-VL rope index scaling." - }, - { - "left": "issue:43577", - "right": "issue:44554", - "accept": false, - "reason": "Blip2 dtype propagation is unrelated to the MPS attention correctness issue." + "reason": "BigBird special-token registration and DiaConfig token-ID loss are different tokenizer/config bugs." } ] }, @@ -5585,32 +5625,33 @@ "issue:44451", "issue:44464", "issue:44479", + "issue:44554", "issue:44610", "issue:44625", "issue:44821", "issue:44849", "issue:44913", - "issue:44964", - "issue:44977", "issue:45084", "issue:45200", "issue:45229", - "issue:45357", - "issue:45405" + "issue:45325", + "issue:45357" ], "soft_pairs": [ + "issue:43329|issue:45325", "issue:43531|issue:43742", + "issue:43577|issue:44554", "issue:44479|issue:44625", "issue:43550|issue:44361", - "issue:43012|issue:43577", "issue:41720|issue:44464", + "issue:43012|issue:43577", "issue:43504|issue:43720", "issue:43504|issue:43550", "issue:32090|issue:33666", - "issue:43262|issue:45084", "issue:32090|issue:34567", - "issue:43278|issue:43688", + "issue:43262|issue:45084", "issue:43504|issue:44610", + "issue:43278|issue:43688", "issue:43452|issue:44451", "issue:45200|issue:45229", "issue:44849|issue:45357", @@ -5621,237 +5662,228 @@ "issue:33666|issue:36010", "issue:43742|issue:44821", "issue:43525|issue:43572", - "issue:43688|issue:43873", "issue:43720|issue:43927", "issue:43606|issue:43927", + "issue:43688|issue:43873", "issue:43278|issue:43844", "issue:43257|issue:43299", "issue:43720|issue:44361", - "issue:39692|issue:43493", - "issue:43540|issue:44977", - "issue:44964|issue:45405", - "issue:43329|issue:43606" + "issue:39692|issue:43493" ], - "prepared_review_unit_hash": "a3982faf02b5512bf0b5936fc1f88c3d1d5cf78ecec081a85bfcd570f4c2c0a1", + "prepared_review_unit_hash": "7d4c231fc819e85be1cfcb4fbce56b25a80d59f99444578850d22187cb606d99", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37248, - "estimated_input_tokens": 9312, - "estimated_eval_tokens": 18880 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36444, + "estimated_input_tokens": 9111, + "estimated_eval_tokens": 18478 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous: most items are unrelated bugs spanning training, loading, multimodal/video, config serialization, and dtype/offload behavior. The soft pairs are only superficially similar, so I would not collapse any of them as duplicates.", - "confidence": 0.91, - "canonical_issue_reason": "No canonical issue stands out: the cluster does not form a single underlying bug or change, and none of the soft pairs look like the same concrete problem.", + "summary": "This cluster is heterogeneous: the soft links mostly connect superficially similar bug reports (config/token-ID regressions, multimodal video issues, dtype/offload problems) rather than the same underlying defect. I reject all soft edges. If one issue must be used as a representative, `issue:43525` is the cleanest self-contained regression report in the set.", + "confidence": 0.83, + "canonical_issue_reason": "No true canonical duplicate stands out; if forced to pick one representative, `issue:43525` is the most concrete and self-contained config regression (`pad_token_id` missing on `Llama4Config`).", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a clear global representative because the items cover many distinct code paths and model families.", + "best_issue_reason": "`issue:43525` is the strongest representative of the cluster\u2019s small config-regression subset: clear symptom, narrow scope, and easy to understand impact.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43329", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen VL/video-related, but one is an undefined-variable bug in multimodal token counting and the other is a rope/position-id scaling regression; different code paths." + }, { "left": "issue:43531", "right": "issue:43742", "accept": false, - "reason": "Different failures: Qwen3-MoE sliding window behavior vs a MobileLLM loading key error; only broad loading/runtime similarity." + "reason": "Sliding-window behavior in Qwen3-MoE and a MobileLLM key error are unrelated model-loading/runtime bugs." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "One is a training/eval dtype mismatch, the other is an MPS attention shape/correctness issue; not the same defect." }, { "left": "issue:44479", "right": "issue:44625", "accept": false, - "reason": "Video-input regression across Qwen VL variants vs Qwen3.5 num_labels propagation; distinct subsystems and symptoms." + "reason": "A video-input regression across Qwen VL variants is distinct from `num_labels` propagation in Qwen3.5." }, { "left": "issue:43550", "right": "issue:44361", "accept": false, - "reason": "torch.compile + SDPA failure in Bamba vs tokenizer AttributeError in MLukeTokenizer; unrelated bugs." + "reason": "Bamba `torch.compile`/SDPA failure and MLuke tokenizer attribute errors are different subsystems and symptoms." }, { - "left": "issue:43012", - "right": "issue:43577", + "left": "issue:41720", + "right": "issue:44464", "accept": false, - "reason": "One is a PyTorch warning during bf16 compilation, the other is dtype not propagating on Blip2 load; not the same defect." + "reason": "CUDA assert with auto device mapping is not the same as inconsistent outputs from compiled chunked generation." }, { - "left": "issue:41720", - "right": "issue:44464", + "left": "issue:43012", + "right": "issue:43577", "accept": false, - "reason": "CUDA assert with auto device mapping for Qwen3 vs chunked generation inconsistency under compiled forward; different code paths." + "reason": "Both mention dtype, but one is a PyTorch warning during compile and the other is a Blip2 loading dtype bug." }, { "left": "issue:43504", "right": "issue:43720", "accept": false, - "reason": "Legacy field in Beit semantic segmentation loading vs packed-weight unpacking in BitNet accelerate loading; different model/loading bugs." + "reason": "Legacy-field loading failure in BEiT and unpacking packed weights in BitNet are different loading bugs." }, { "left": "issue:43504", "right": "issue:43550", "accept": false, - "reason": "Completely different models and issues: Beit preset loading vs Bamba compile/SDPA behavior." + "reason": "Unrelated model-loading bug vs compile/SDPA runtime failure." }, { "left": "issue:32090", "right": "issue:33666", "accept": false, - "reason": "Trainer _gpu_broadcast_one NoneType error vs Qwen2-VL multi-GPU training; too broad to be the same bug." - }, - { - "left": "issue:43262", - "right": "issue:45084", - "accept": false, - "reason": "Audio chat-template sample-rate defaulting vs template compilation error; unrelated feature areas." + "reason": "Trainer broadcast error and Qwen2-VL multi-GPU training are too broad to be the same concrete bug." }, { "left": "issue:32090", "right": "issue:34567", "accept": false, - "reason": "Trainer broadcast TypeError vs TrainerState token counter not updating; different training issues." + "reason": "Different Trainer issues: GPU broadcast type error vs token-count bookkeeping not updating." }, { - "left": "issue:43278", - "right": "issue:43688", + "left": "issue:43262", + "right": "issue:45084", "accept": false, - "reason": "BF16-to-FP32 dtype drift during evaluate vs auxiliary-loss normalization error in MoE models; no shared concrete bug." + "reason": "Audio chat-template sampling-rate default and template-node compilation errors are unrelated." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "Pretrained-load legacy field issue vs processor output size mismatch; different models and failure modes." + "reason": "BEiT legacy config loading and OmDet-Turbo processor/model size mismatch are distinct bugs." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Training/eval dtype drift is unrelated to auxiliary-loss normalization." }, { "left": "issue:43452", "right": "issue:44451", "accept": false, - "reason": "Generic gguf_file loading breakage vs a specific ScandiBERT load failure; too little evidence they share the same root cause." + "reason": "A `gguf_file` path handling bug and a specific model load failure for ScandiBERT are not the same issue." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "mm_token_type_ids defaulting issue for text-only fine-tuning vs multi-GPU inference OOM; unrelated." + "reason": "Gemma4 text-only token-type defaults and multi-GPU CUDA OOM are different problems." }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "output_hidden_states bug in Qwen3.5 vs save_pretrained saving wrong visual encoder keys; different code paths." + "reason": "`output_hidden_states` behavior and incorrect visual-encoder keys in `save_pretrained` are separate regressions." }, { "left": "issue:43334", "right": "issue:43572", "accept": false, - "reason": "Both are config-attribute errors, but one is pad_token_id in Qwen3-VL and the other is pad_token_idx in StableLm; not the same bug." + "reason": "Both are missing token-id fields in configs, but they affect different models/attributes and do not share a concrete code path." }, { "left": "issue:43756", "right": "issue:44913", "accept": false, - "reason": "Smollm3 RoPE-layer count mismatch vs GPTNeoX rotary_pct reload bug; both touch config semantics but are distinct." + "reason": "RoPE-layer count mismatch and `rotary_pct` reset on reload are different configuration bugs." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 implementation discrepancy vs pipeline translation-task error messaging; completely unrelated." + "reason": "SigLIP2 implementation discrepancy and a pipeline translation error message are unrelated." }, { "left": "issue:43540", "right": "issue:43716", "accept": false, - "reason": "Qwen3OmniMoe video-processing ValueError vs Mistral-3 image-preprocessor dtype mismatch; different modalities and causes." + "reason": "Qwen3OmniMoe video processing failure and Mistral-3 image-preprocessor dtype mismatch are different multimodal bugs." }, { "left": "issue:33666", "right": "issue:36010", "accept": false, - "reason": "Multi-GPU Qwen2-VL training vs GenerationMixin import failure; no common concrete bug." + "reason": "Qwen2-VL multi-GPU training and a `GenerationMixin` import error have no concrete overlap." }, { "left": "issue:43742", "right": "issue:44821", "accept": false, - "reason": "MobileLLM loading key error vs AutoImageProcessor-from-URL loading issue; both are load failures but not the same one." + "reason": "MobileLLM key error and loading `AutoImageProcessor` from URL are unrelated loading failures." }, { "left": "issue:43525", "right": "issue:43572", "accept": false, - "reason": "Llama4 missing pad_token_id vs StableLm missing pad_token_idx; similar shape, but different model families and attributes." - }, - { - "left": "issue:43688", - "right": "issue:43873", - "accept": false, - "reason": "Auxiliary-loss normalization bug in MoE models vs quantization/offloading problem; unrelated." + "reason": "Both involve missing token-id fields, but the missing attribute and model/config context differ; not the same underlying bug." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "BitNet packed-weight handling vs DiaConfig token ID persistence; both loading-related, but different state and code paths." + "reason": "BitNet packed-weight unpacking on accelerate load and DiaConfig token-ID persistence are separate failures." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch vs DiaConfig save/load token ID loss; not the same failure." + "reason": "CPU offload device mismatch is unrelated to save/load loss of custom token IDs." + }, + { + "left": "issue:43688", + "right": "issue:43873", + "accept": false, + "reason": "Aux-loss normalization and quantized offloading are different behaviors in different code paths." }, { "left": "issue:43278", "right": "issue:43844", "accept": false, - "reason": "Training/evaluate dtype drift vs gradient explosion under ZeRO-3; different symptoms and mechanisms." + "reason": "Embedding dtype drift and ZeRO-3 gradient amplification are not the same bug." }, { "left": "issue:43257", "right": "issue:43299", "accept": false, - "reason": "Both involve Qwen3 MoE loading, but one is accelerate+deepspeed weight conversion and the other is a transformers-dev load regression for Qwen3VL MoE; not clearly the same bug." + "reason": "Both concern Qwen3 MoE loading, but one is accelerate/deepspeed weight conversion and the other is a transformers-dev regression for Qwen3VL MoE; too different to merge." }, { "left": "issue:43720", "right": "issue:44361", "accept": false, - "reason": "Packed-weight loading bug in BitNet vs MLukeTokenizer task AttributeError; no overlap in root cause." + "reason": "BitNet weight unpacking and MLuke tokenizer attribute errors are unrelated." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "SigLIP2 docs/example issues vs SigLIP2 implementation discrepancy with JAX; one is documentation/example breakage, the other is model parity." - }, - { - "left": "issue:43540", - "right": "issue:44977", - "accept": false, - "reason": "Qwen3OmniMoe video-input failure vs Qwen3.5 flash-attention generation bug; different model paths." - }, - { - "left": "issue:44964", - "right": "issue:45405", - "accept": false, - "reason": "Phi-4 multimodal load failure vs peft version bump to an unreleased release; unrelated." - }, - { - "left": "issue:43329", - "right": "issue:43606", - "accept": false, - "reason": "Undefined video token helper in multimodal token counting vs bark-small CPU offload device mismatch; not the same bug." + "reason": "SigLIP2 docs-example failures and SigLIP2 HF-vs-JAX numerical discrepancy are different issues, not the same underlying bug." } ] }, @@ -5884,6 +5916,7 @@ "issue:43257", "issue:43278", "issue:43295", + "issue:43329", "issue:43493", "issue:43504", "issue:43540", @@ -5915,15 +5948,18 @@ "issue:44734", "issue:44792", "issue:44843", + "issue:44964", "issue:44977", "issue:45084", - "issue:45200", "issue:45290", "issue:45405", "issue:45446", "issue:45520" ], "soft_pairs": [ + "issue:44964|issue:45405", + "issue:43540|issue:44977", + "issue:43329|issue:43606", "issue:33666|issue:36331", "issue:44488|issue:44843", "issue:43493|issue:43844", @@ -5934,8 +5970,8 @@ "issue:44521|issue:45290", "issue:44423|issue:44734", "issue:33357|issue:33666", - "issue:39692|issue:43295", "issue:43824|issue:44589", + "issue:39692|issue:43295", "issue:43994|issue:44493", "issue:43504|issue:43606", "issue:43927|issue:44361", @@ -5949,199 +5985,210 @@ "issue:15354|issue:38175", "issue:44554|issue:45446", "issue:43582|issue:44977", - "issue:45084|issue:45520", - "issue:43540|issue:45200" + "issue:45084|issue:45520" ], - "prepared_review_unit_hash": "3db9abe3ceee849d8935581f74d3002afe8107a155a89ee84d98a10020b999c8", + "prepared_review_unit_hash": "0fdfd3cb3c264f3a2a6c12122b9585ebab41b938b72bf1c288d2b14c372449c5", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 27, - "serialized_chars": 35428, - "estimated_input_tokens": 8857, - "estimated_eval_tokens": 17970 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36681, + "estimated_input_tokens": 9171, + "estimated_eval_tokens": 18598 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab bag of unrelated standalone bugs across different models, runtimes, and subsystems. I don\u2019t see a safe duplicate anchor or any PR-style merge candidate.", - "confidence": 0.98, - "canonical_issue_reason": null, + "summary": "This cluster is a loose semantic neighborhood of Transformer bug reports, mostly around multimodal/model-loading regressions, serving, tokenizer/chat-template behavior, and a few unrelated training/runtime issues. I do not see any soft pair that is clearly the same underlying bug/change, so I would not merge any of them as duplicates.", + "confidence": 0.88, + "canonical_issue_reason": "issue:44521 is the broadest and most central active report in this set: multimodal chat-template handling with downstream references, which makes it the best representative anchor even though the cluster overall is heterogeneous.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:44521 is the best single issue to keep as a representative because it targets a high-signal multimodal API path and has the most visible downstream interest in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "One is a Phi-4 multimodal model load failure; the other is a PEFT version bump/regression. Different problems." + }, + { + "left": "issue:43540", + "right": "issue:44977", + "accept": false, + "reason": "Different models and failure modes: Qwen3OmniMoe video processing vs Qwen3.5 flash-attention generation." + }, + { + "left": "issue:43329", + "right": "issue:43606", + "accept": false, + "reason": "Undefined multimodal-token helper bug vs CPU offload device mismatch. Not the same code path." + }, { "left": "issue:33666", "right": "issue:36331", "accept": false, - "reason": "Different failures: Qwen2-VL multi-GPU training vs a Trainer API kwarg regression." + "reason": "Multi-GPU training for Qwen2-VL vs a CustomTrainer signature regression. Unrelated." }, { "left": "issue:44488", "right": "issue:44843", "accept": false, - "reason": "Both involve model loading, but one is a specific model load failure and the other is an offline-mode tokenizer patch bug." + "reason": "Both are loading-related, but one is a specific model failure and the other is an offline-mode hub call regression. Too broad to merge." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "Unrelated symptoms: SigLIP2 implementation discrepancy vs ZeRO-3 gradient explosion." + "reason": "SigLIP2 implementation discrepancy vs a DeepSpeed/ZeRO-3 gradient issue. Different subsystems and symptoms." }, { "left": "issue:44589", "right": "issue:45520", "accept": false, - "reason": "Completely different areas: float8 storage lookup vs flash_attn import on Python 3.13." + "reason": "Float8 storage lookup failure vs flash_attn import error on Python 3.13. Unrelated runtime issues." }, { "left": "issue:43756", "right": "issue:44060", "accept": false, - "reason": "Both are Qwen-related model issues, but one is RoPE layer dropping and the other is a tied-weights warning bug." + "reason": "Smollm3 RoPE-layer mismatch vs Qwen3-Next tied-weights warning. Different model internals." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Different code paths: PEFT/local checkpoint loading vs serialization of tied weights depending on device." + "reason": "Custom model initialization regression vs tied-weight serialization inconsistency. Both v5-related, but not the same bug." }, { "left": "issue:43688", "right": "issue:43746", "accept": false, - "reason": "Distinct bugs in different models: auxiliary loss normalization vs GraniteSpeech local checkpoint loading." + "reason": "Aux-loss normalization in MoE models vs GraniteSpeech PEFT checkpoint loading. Different affected paths." }, { "left": "issue:44521", "right": "issue:45290", "accept": false, - "reason": "Same API family, but one is wrong assistant masks for multimodal inputs and the other is a crash on tool-call messages without content." + "reason": "Both involve apply_chat_template, but one is multimodal assistant masks and the other is tool-call content handling. Not the same defect." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both hit serve/multimodal flows, but one is a '.to' AttributeError and the other is KV-cache continuation indexing." + "reason": "Serve crashes in two different routes: multimodal continuous batching vs KV-cache continuation indexing." }, { "left": "issue:33357", "right": "issue:33666", "accept": false, - "reason": "Different models and failure modes: Mac bus error on CLIP vs Qwen2-VL training issue." + "reason": "MacOS bus error on a CLIP model vs Qwen2-VL multi-GPU training. No shared underlying bug." }, { - "left": "issue:39692", - "right": "issue:43295", + "left": "issue:43824", + "right": "issue:44589", "accept": false, - "reason": "Both mention multimodal examples, but one is a docs/example mismatch and the other is a regression in processor/tokenizer behavior." + "reason": "ImportError for a missing model class vs a float8 storage TypeError. Different failure layers." }, { - "left": "issue:43824", - "right": "issue:44589", + "left": "issue:39692", + "right": "issue:43295", "accept": false, - "reason": "Unrelated: missing Qwen2.5-VL import vs missing Float8 storage object." + "reason": "Docs example errors/quantization mismatch vs a regression in processor.tokenizer access and image passing. Related area, but not the same issue." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "Different issues: SigLIP2 bad outputs vs a broad position-id key warning regression." + "reason": "SigLIP2 bad outputs vs position-id key warnings across many models. Different observable bugs." }, { "left": "issue:43504", "right": "issue:43606", "accept": false, - "reason": "Different bugs in different models: legacy field load failure vs CPU offload device mismatch." + "reason": "Legacy field blocks BEiT semantic segmentation loading vs bark-small CPU offload mismatch. Unrelated." }, { "left": "issue:43927", "right": "issue:44361", "accept": false, - "reason": "Separate tokenizer/config bugs with different failure points and models." + "reason": "Config token-id persistence bug vs tokenizer AttributeError in tasks. Different objects and root causes." }, { "left": "issue:43504", "right": "issue:43653", "accept": false, - "reason": "No shared root cause: model preset loading vs BigBird tokenizer special-token registration." + "reason": "BEiT preset loading issue vs BigBirdTokenizer special-token registration bug. Not the same defect." }, { "left": "issue:43575", "right": "issue:45405", "accept": false, - "reason": "One is TP/OOM on a large model; the other is a dependency version mismatch." + "reason": "TP OOM when loading a large Qwen model vs a PEFT minimum-version bump. Completely different." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "Different init/config regressions: missing pad_token_idx vs unexpected _is_hf_initialized argument." + "reason": "Missing pad_token_idx in one config vs an init_empty_weights signature regression. Both 5.0-era loading issues, but not the same bug." }, { "left": "issue:43278", "right": "issue:43701", "accept": false, - "reason": "Unrelated training/eval dtype change vs checkpoint resume key mismatch." + "reason": "Embedding dtype drift in evaluate vs resume_from_checkpoint key mismatch. Different training lifecycle problems." }, { "left": "issue:44792", "right": "issue:45405", "accept": false, - "reason": "Janus image-generation test failure is unrelated to the PEFT version bump." + "reason": "Janus image-generation test failure vs PEFT version requirement. No overlap." }, { "left": "issue:43257", "right": "issue:45405", "accept": false, - "reason": "Qwen3 MoE weight conversion/loading issue is unrelated to a release version pin problem." + "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed vs a dependency version bump. Unrelated." }, { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "One is a real device mismatch bug; the other is a stale GraniteSpeech CI test." + "reason": "Both mention device-related CI behavior, but one is a real CPU-offload runtime mismatch and the other is a stale GraniteSpeech test. Not the same issue." }, { "left": "issue:15354", "right": "issue:38175", "accept": false, - "reason": "TorchScript GeneratorExp export failure vs unexpected zero probabilities in SigLIP2 are unrelated." + "reason": "TorchScript GeneratorExp export failure vs unexpected zero probabilities in SigLIP2. Different eras and code paths." }, { "left": "issue:44554", "right": "issue:45446", "accept": false, - "reason": "Different attention-related bugs with different trigger conditions and APIs." + "reason": "MPS attention correctness issue vs a PyTorch version gate bug in flex_attention import. Different layers." }, { "left": "issue:43582", "right": "issue:44977", "accept": false, - "reason": "AppleSilicon allocator TypeError and Qwen3.5 flash-attention generation issues are unrelated." + "reason": "Apple Silicon allocator TypeError vs Qwen3.5 flash-attention generation bug. Unrelated." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Template compilation error is unrelated to the flash_attn import-key error." - }, - { - "left": "issue:43540", - "right": "issue:45200", - "accept": false, - "reason": "Different multimodal bugs: video processing ValueError vs missing mm_token_type_ids for text-only fine-tuning." + "reason": "Template compilation TypeError vs flash_attn import KeyError. Different failure mechanisms." } ] }, @@ -6178,6 +6225,7 @@ "issue:43493", "issue:43525", "issue:43526", + "issue:43540", "issue:43572", "issue:43575", "issue:43701", @@ -6187,7 +6235,6 @@ "issue:43873", "issue:43957", "issue:43975", - "issue:43994", "issue:44315", "issue:44442", "issue:44464", @@ -6215,6 +6262,7 @@ "issue:45563" ], "soft_pairs": [ + "issue:43540|issue:45200", "issue:45072|issue:45198", "issue:43844|issue:44964", "issue:43232|issue:44849", @@ -6240,94 +6288,99 @@ "issue:43572|issue:44987", "issue:44442|issue:45198", "issue:43295|issue:43873", - "issue:43824|issue:43975", "issue:43572|issue:44589", + "issue:43824|issue:43975", "issue:43278|issue:43856", - "issue:38175|issue:45412", - "issue:43295|issue:43994" + "issue:38175|issue:45412" ], - "prepared_review_unit_hash": "af5d21768b7dc5b40b69e75af71b1d0fb4b06fd35d22f833e2ccf6bd970f5a39", + "prepared_review_unit_hash": "523f235e5bf0b505ea1aac431f30765550e18e75cbacc32394f7428d9b2248d7", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36891, - "estimated_input_tokens": 9223, - "estimated_eval_tokens": 18702 + "serialized_chars": 36866, + "estimated_input_tokens": 9217, + "estimated_eval_tokens": 18690 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a heterogeneous mix of unrelated Transformers bugs (imports, configs, tokenizers, generation, processors, training/runtime issues). I don\u2019t see a single duplicate thread to consolidate, so all soft edges are rejected.", - "confidence": 0.96, - "canonical_issue_reason": "No canonical issue: the items span different code paths, models, and failure modes, so there isn\u2019t one safe representative duplicate.", + "summary": "No strong duplicate cluster emerges from these items; they span unrelated bugs across tokenizers, multimodal processors, training, generation, and model loading. The most broadly representative issue is the open multimodal assistant-mask bug.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is specific, reproducible, and still open, with clear multimodal apply_chat_template failure semantics and active discussion/inbound references.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue: any single choice would collapse distinct bugs into one and misrepresent the cluster.", + "best_issue_reason": "issue:44521 is the best representative issue because it is concrete, currently open, and likely to benefit from an upstream fix without being overly model-specific.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Both involve multimodal inputs, but the concrete failures differ: video input processing ValueError vs missing mm_token_type_ids defaulting. Different code paths and fixes." + }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "Different failures: dtype mismatch in inference vs save_pretrained/tokenization on Wav2Vec2." + "reason": "Unrelated bugs in different subsystems/models: dtype mismatch in inference vs save_pretrained/tokenization failure." }, { "left": "issue:43844", "right": "issue:44964", "accept": false, - "reason": "Training/ZeRO gradient behavior is unrelated to a model-loading failure." + "reason": "Different problems: ZeRO-3 gradient inflation during training vs model loading failure for Phi-4 multimodal." }, { "left": "issue:43232", "right": "issue:44849", "accept": false, - "reason": "Different generation/model-output bugs with no shared concrete code path." + "reason": "Generation kwargs update after sync_gpus is a different bug from Qwen3.5 output_hidden_states handling." }, { "left": "issue:39401", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer issues, but they affect different models and symptoms (offset mapping vs missing BOS/EOS)." + "reason": "Both are tokenizer bugs, but they affect different models and behaviors: offset_mapping vs special-token insertion." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "ImportError for a missing helper is unrelated to meta-device loading breakage." + "reason": "ImportError for a missing flash-attention helper is not the same as meta-device loading breakage." }, { "left": "issue:44315", "right": "issue:44849", "accept": false, - "reason": "Liger Kernel application under model_init is unrelated to Qwen3.5 hidden-states behavior." + "reason": "Liger Kernel application in model_init is unrelated to output_hidden_states behavior." }, { "left": "issue:44554", "right": "issue:45356", "accept": false, - "reason": "MPS attention correctness and tokenizer codec handling are unrelated bugs." + "reason": "Different domains entirely: MPS attention correctness vs tokenizer regression in Kimi-K2.5." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Both involve processors, but one is a regression around processor/tokenizer access and the other is an image-size mismatch." + "reason": "Both involve processors, but one is a regression around processor.tokenizer/image passing, the other is an image-size mismatch." }, { "left": "issue:43334", "right": "issue:43525", "accept": false, - "reason": "Similar symptom, but different config classes and likely different model integrations." + "reason": "Similar symptom shape, but different model configs and likely separate missing-field fixes; not enough evidence for one underlying bug." }, { "left": "issue:43493", @@ -6339,121 +6392,115 @@ "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Both are multimodal-related, but one is token-type defaults and the other is rope index temporal scaling." + "reason": "Different Qwen multimodal issues: mm_token_type_ids defaulting vs still-image temporal rope scaling." }, { "left": "issue:44898", "right": "issue:45412", "accept": false, - "reason": "Image resolution/interpolation failure is unrelated to a memory-release problem." + "reason": "Perceiver resolution handling is unrelated to RT-DETR memory release on deletion." }, { "left": "issue:44466", "right": "issue:45127", "accept": false, - "reason": "Serialization inconsistency and LoRA merge collapse are different bugs in different code paths." + "reason": "Serialization of tied lm_head weights is unrelated to LoRA merge collapse after vocab extension." }, { "left": "issue:43873", "right": "issue:44464", "accept": false, - "reason": "Quantization offloading and compiled-forward chunked generation are unrelated." + "reason": "Quantization/offloading behavior is unrelated to chunked generation inconsistencies under compiled forward." }, { "left": "issue:40990", "right": "issue:44625", "accept": false, - "reason": "Perplexity regression on one model is unrelated to num_labels propagation on another." + "reason": "Perplexity regression and num_labels propagation are different bugs with different failure modes." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "Janus image-generation test failure is unrelated to a stale warning in paged generate()." + "reason": "Failed janus image-generation test is not the same as a stale warning in paged generate()." }, { "left": "issue:43388", "right": "issue:43526", "accept": false, - "reason": "Both touch labels, but in different subsystems and with different failure mechanisms." + "reason": "Both touch labels, but one is trainer metrics batching and the other is BeitImageProcessorFast reducing labels incorrectly." }, { "left": "issue:44514", "right": "issue:44521", "accept": false, - "reason": "Same processor family, but one crashes on batched padding=False while the other produces zero assistant masks." + "reason": "Both are in apply_chat_template, but one crashes on batched padding=False while the other produces zero assistant masks; different concrete failures." }, { "left": "issue:43844", "right": "issue:44936", "accept": false, - "reason": "DeepSpeed/ZeRO gradient behavior is unrelated to trainer.evaluate() after train()." + "reason": "Training instability with DeepSpeed is unrelated to evaluate() failing after train()." }, { "left": "issue:43575", "right": "issue:44464", "accept": false, - "reason": "TP OOM during model load is unrelated to chunked generation inconsistency." + "reason": "Tensor-parallel OOM during model load is unrelated to compiled-forward chunked generation output differences." }, { "left": "issue:37428", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different modules/models." + "reason": "Both are ImportErrors, but for different missing symbols and different affected APIs/models." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor access regression and batched video shape bug are different issues." + "reason": "Processor.tokenizer regression is not the same as batched video processor shape handling." }, { "left": "issue:43572", "right": "issue:44987", "accept": false, - "reason": "Missing pad_token_idx in one config is unrelated to a separate model-loading failure." + "reason": "Missing pad_token_idx in StableLmConfig is different from a broad loading failure for physical-intelligence/fast." }, { "left": "issue:44442", "right": "issue:45198", "accept": false, - "reason": "Different audio tokenization problems with different failure modes." + "reason": "Different tokenizer/model failures: FastSpeech2ConformerTokenizer loading vs Wav2Vec2 save/tokenization issues." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "Processor regression and quantization offloading are unrelated." + "reason": "Custom processor API regression is unrelated to offloading not working with quantization." }, { - "left": "issue:43824", - "right": "issue:43975", + "left": "issue:43572", + "right": "issue:44589", "accept": false, - "reason": "Importing a missing class and incorrect detokenization are unrelated bugs." + "reason": "Missing config attribute vs Float8 storage lookup failure are unrelated." }, { - "left": "issue:43572", - "right": "issue:44589", + "left": "issue:43824", + "right": "issue:43975", "accept": false, - "reason": "Config attribute missing and Float8 storage lookup failure are different runtime issues." + "reason": "Importing a missing class and incorrect detokenization are separate issues." }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "Dtype drift between train/eval and MoE memory inefficiency are unrelated." + "reason": "BF16-to-FP32 evaluation dtype shift is unrelated to Qwen3 MoE memory usage during training." }, { "left": "issue:38175", "right": "issue:45412", "accept": false, - "reason": "Zero-probability inference bug is unrelated to a model memory leak." - }, - { - "left": "issue:43295", - "right": "issue:43994", - "accept": false, - "reason": "Different models and symptoms; processor regression does not match SigLIP2 nonsensical outputs." + "reason": "Unexpected zero probabilities in SigLIP2 and RT-DETR memory release are unrelated." } ] }, @@ -6485,6 +6532,7 @@ "issue:42915", "issue:43262", "issue:43278", + "issue:43295", "issue:43335", "issue:43425", "issue:43450", @@ -6501,6 +6549,7 @@ "issue:43827", "issue:43901", "issue:43931", + "issue:43994", "issue:44112", "issue:44351", "issue:44360", @@ -6513,13 +6562,11 @@ "issue:44743", "issue:44779", "issue:44849", - "issue:44855", "issue:44857", "issue:44991", "issue:45003", "issue:45083", "issue:45127", - "issue:45200", "issue:45245", "issue:45290", "issue:45310", @@ -6527,6 +6574,7 @@ "issue:45478" ], "soft_pairs": [ + "issue:43295|issue:43994", "issue:44360|issue:44561", "issue:38175|issue:43493", "issue:43827|issue:43901", @@ -6537,8 +6585,8 @@ "issue:44568|issue:44991", "issue:45127|issue:45245", "issue:45290|issue:45381", - "issue:43819|issue:45083", "issue:43450|issue:43575", + "issue:43819|issue:45083", "issue:44462|issue:44561", "issue:44561|issue:44779", "issue:44610|issue:44857", @@ -6551,244 +6599,229 @@ "issue:45310|issue:45478", "issue:37428|issue:43425", "issue:43611|issue:43931", - "issue:39692|issue:43526", "issue:43335|issue:43550", "issue:43819|issue:44493", + "issue:39692|issue:43526", "issue:43450|issue:43526", - "issue:37428|issue:44351", - "issue:44855|issue:45083", - "issue:39692|issue:44493", - "issue:45200|issue:45381" + "issue:37428|issue:44351" ], - "prepared_review_unit_hash": "0b4561722ffb19a0c9ebe3f6d8e3bda9650770da0991d9edfb8baaee2e76a8a7", + "prepared_review_unit_hash": "83e2a2b27d99a538056955caeceb6debbf55197175c0c5ac0608d60878adf4b9", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37288, - "estimated_input_tokens": 9322, - "estimated_eval_tokens": 18900 + "soft_pair_count": 30, + "serialized_chars": 36670, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most pairs are unrelated regressions across different models or subsystems. The only near-duplicate pair is the Qwen3.5 MoE from_pretrained reports: same title and very similar symptom, so 45310 is the best canonical issue and 45478 looks like a likely duplicate.", - "confidence": 0.89, - "canonical_issue_reason": "issue:45310 is the best canonical issue because it appears to be the earlier report of the same Qwen3.5 MoE from_pretrained failure; issue:45478 has the same title and reads like a likely duplicate rather than a distinct bug.", + "summary": "This cluster is mostly a loose mix of unrelated Transformers issues across models, tokenizers, processors, docs, and runtime regressions. The only clear duplicate pair is the two identical Qwen3.5 Moe from_pretrained reports; everything else looks like separate bugs or feature regressions.", + "confidence": 0.96, + "canonical_issue_reason": "issue:45310 is the best canonical issue because it is the earliest clear report of the only exact duplicate pair in the cluster, with the same Qwen3.5 Moe from_pretrained failure later repeated in issue:45478.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the strongest issue candidate overall because it best represents the likely duplicate Qwen3.5 MoE from_pretrained reports and has the earlier timeline.", + "best_issue_reason": "issue:45310 is the strongest representative issue for this cluster: it is concrete, specific, and has a true duplicate counterpart, whereas most other items are unrelated single-bug reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:43994", + "accept": false, + "reason": "Different failures: processor.tokenizer/image handling regression vs SigLIP2 nonsensical outputs." + }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "Different bugs: DSA indexer activation detail versus a v5 compatibility break from removing is_torch_fx_available." + "reason": "Unrelated topics: DSA indexer ReLU discussion vs removal of is_torch_fx_available breaking trust_remote_code." }, { "left": "issue:38175", "right": "issue:43493", "accept": false, - "reason": "Both mention SigLIP2, but the symptoms differ: zero probabilities versus HF-vs-JAX implementation mismatch." + "reason": "Same model family, but different concrete symptoms; not enough to treat as one bug." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs/pipeline-related, but they describe different documentation issues and different behavior changes." + "reason": "Both are docs-related, but they cover different APIs and different problems." }, { "left": "issue:44849", "right": "issue:45003", "accept": false, - "reason": "Unrelated: Qwen3.5 hidden-state behavior versus unsafe sys.modules access in modeling_utils." + "reason": "Qwen3.5 hidden-states bug is unrelated to sys.modules access in modeling_utils." }, { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "Different areas: tokenizer special-token registration versus a GraniteSpeech CI device-override test." + "reason": "Different components and failures: tokenizer special-token bug vs GraniteSpeech CI device override test." }, { "left": "issue:42915", "right": "issue:43824", "accept": false, - "reason": "Different failures: Qwen3Moe FineGrainedFP8Config versus a Qwen2.5-VL import error." + "reason": "FP8 config failure in Qwen3Moe is unrelated to Qwen2.5-VL import error." }, { "left": "issue:43450", "right": "issue:43493", "accept": false, - "reason": "Video processor shape bug versus SigLIP2 implementation discrepancy; no shared concrete code path." + "reason": "Batched video shape bug is not the same as SigLIP2 implementation discrepancy." }, { "left": "issue:44568", "right": "issue:44991", "accept": false, - "reason": "Both concern tokenizer behavior in v5, but they affect different models and different failure modes." + "reason": "Both concern tokenizers in v5, but they are different models with different failure modes." }, { "left": "issue:45127", "right": "issue:45245", "accept": false, - "reason": "Unrelated issues: LoRA/extended-vocab collapse versus a category-count runtime limit." + "reason": "LoRA/tied embeddings collapse and category-count RuntimeError are unrelated bugs." }, { "left": "issue:45290", "right": "issue:45381", "accept": false, - "reason": "Different chat-template/tool-call crash versus a Qwen2.5-VL video position-id issue." + "reason": "Chat template tool-call crash and Qwen2.5-VL video position ids are different code paths." }, { - "left": "issue:43819", - "right": "issue:45083", + "left": "issue:43450", + "right": "issue:43575", "accept": false, - "reason": "Different model paths and symptoms: DAC latent mismatch versus qwen3_omni_moe helper behavior." + "reason": "Video processor shape bug and tp OOM on Qwen2-57B are unrelated." }, { - "left": "issue:43450", - "right": "issue:43575", + "left": "issue:43819", + "right": "issue:45083", "accept": false, - "reason": "Video processor shape bug and Qwen2-57B tensor-parallel OOM are unrelated." + "reason": "DAC from_latents mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." }, { "left": "issue:44462", "right": "issue:44561", "accept": false, - "reason": "Tokenizer.json handling and is_torch_fx_available removal are different regressions." + "reason": "AutoTokenizer repository-loading issue is not the same as the v5 trust_remote_code regression." }, { "left": "issue:44561", "right": "issue:44779", "accept": false, - "reason": "Different failures: trust_remote_code compatibility versus DeepSeek tokenizer correctness." + "reason": "Both are v5-related, but one is API removal and the other is Deepseek tokenizer output regression." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "Input-size mismatch in OmDet-Turbo versus an AMP/CUDA crash in LwDetrImageLoss." + "reason": "OmDet-Turbo input-size mismatch is unrelated to LwDetrImageLoss AMP/CUDA crashing." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio chat-template sampling-rate default versus video batched-shape output; different subsystems." + "reason": "Audio chat-template sample-rate default bug and video batch-shape bug are different processor issues." }, { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "Different bugs and code paths: DAC forward mismatch versus a GraniteSpeech CI test issue." + "reason": "DAC latent/STE mismatch is unrelated to GraniteSpeech CI device override flakiness." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Different regressions: Pegasus output drift across versions versus qwen3_5 recurrent-state reset." + "reason": "Pegasus v4/v5 output differences are unrelated to qwen3_5 recurrent-state reset behavior." }, { "left": "issue:43606", "right": "issue:45083", "accept": false, - "reason": "CPU offload device mismatch is unrelated to qwen3_omni_moe helper behavior." + "reason": "CPU offload device mismatch in bark-small is unrelated to qwen3_omni_moe helper behavior." }, { "left": "issue:43278", "right": "issue:43825", "accept": false, - "reason": "Embedding dtype change during evaluate and a pipeline error-message issue are unrelated." + "reason": "Embedding dtype drift in eval and pipeline translation error-message regression are different bugs." }, { "left": "issue:29942", "right": "issue:32090", "accept": false, - "reason": "Both are failures, but Flash Attention 2 tests and a Trainer broadcast type error are different bugs." + "reason": "Flash Attention 2 test failures are unrelated to Trainer _gpu_broadcast_one NoneType errors." }, { "left": "issue:45310", "right": "issue:45478", "accept": true, - "reason": "Same title and same Qwen3.5 MoE from_pretrained failure; 45478 looks like a likely duplicate of 45310." + "reason": "Identical title and same Qwen3.5 Moe from_pretrained failure; these are the same underlying bug." }, { "left": "issue:37428", "right": "issue:43425", "accept": false, - "reason": "ImportError for a flash-attention helper versus a Torch 2.10 compatibility issue." + "reason": "ImportError for flash-attention helper is not the same as torch 2.10 incompatibility." }, { "left": "issue:43611", "right": "issue:43931", "accept": false, - "reason": "Different model-loading failures with different root causes." - }, - { - "left": "issue:39692", - "right": "issue:43526", - "accept": false, - "reason": "SigLIP2 docs/example errors versus BeitImageProcessorFast reduce_labels behavior are unrelated." + "reason": "base_model_prefix loading breakage and Qwen3-VL shape mismatch are unrelated model-loading bugs." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "SwitchTransformers sparse-layer creation bug and Bamba torch.compile SDPA failure are different code paths." + "reason": "SwitchTransformers sparse-layer config bug and Bamba torch.compile/SDPA failure are different issues." }, { "left": "issue:43819", "right": "issue:44493", "accept": false, - "reason": "Different symptoms and subsystems: DAC latent mismatch versus unexpected position-id keys across many models." + "reason": "DAC latent mismatch is unrelated to position-id key regressions across models." }, { - "left": "issue:43450", + "left": "issue:39692", "right": "issue:43526", "accept": false, - "reason": "Different processors and different bugs." - }, - { - "left": "issue:37428", - "right": "issue:44351", - "accept": false, - "reason": "Different import errors for different missing symbols." - }, - { - "left": "issue:44855", - "right": "issue:45083", - "accept": false, - "reason": "Python 3.13 Deberta import parsing issue versus qwen3_omni_moe helper behavior." + "reason": "SigLIP2 docs example errors are unrelated to BeitImageProcessorFast reduce_labels bug." }, { - "left": "issue:39692", - "right": "issue:44493", + "left": "issue:43450", + "right": "issue:43526", "accept": false, - "reason": "Documentation/example errors in SigLIP2 versus a broad position-id key regression; not the same issue." + "reason": "Video batch-shape bug and BeitImageProcessorFast label reduction bug are different processor paths." }, { - "left": "issue:45200", - "right": "issue:45381", + "left": "issue:37428", + "right": "issue:44351", "accept": false, - "reason": "Gemma 4 token-type-id defaulting and Qwen2.5-VL video position ids are separate bugs." + "reason": "Both are import errors, but they involve different symbols and different root causes." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The analysis is mostly conservative, but it overstates the 45310/45478 relationship: the packet shows identical titles and strong similarity, which supports a likely duplicate, but not a confirmed duplicate. Please soften that language (e.g., \"likely duplicate\" / \"best canonical issue\") and avoid claiming certainty beyond the packet evidence." + "accept": true, + "feedback": "Grounded and conservative overall. The summary matches the packet: most items are clearly unrelated by title, and the only clearly supported duplicate pair in the provided data is issue:45310 / issue:45478, which share the same Qwen3.5 Moe from_pretrained error. The soft-edge rejections are appropriately cautious, and the one accepted pair is well supported by identical titles and chronology." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -6814,6 +6847,7 @@ "issue:35141", "issue:37428", "issue:38175", + "issue:39692", "issue:40444", "issue:41720", "issue:43066", @@ -6830,7 +6864,6 @@ "issue:43720", "issue:43723", "issue:43873", - "issue:43976", "issue:44112", "issue:44292", "issue:44448", @@ -6839,6 +6872,7 @@ "issue:44554", "issue:44568", "issue:44779", + "issue:44855", "issue:44857", "issue:44977", "issue:44987", @@ -6848,17 +6882,19 @@ "issue:45229", "issue:45245", "issue:45372", + "issue:45381", "issue:45406", - "issue:45440", - "issue:45459", "issue:45491", "issue:45520", "issue:45563" ], "soft_pairs": [ + "issue:44855|issue:45083", + "issue:45200|issue:45381", + "issue:39692|issue:44493", "issue:43404|issue:45200", - "issue:29127|issue:32090", "issue:43335|issue:43653", + "issue:29127|issue:32090", "issue:44448|issue:44779", "issue:44977|issue:45229", "issue:38175|issue:44493", @@ -6869,15 +6905,15 @@ "issue:43335|issue:45072", "issue:44568|issue:45491", "issue:43720|issue:44112", - "issue:40444|issue:43526", "issue:44448|issue:44977", - "issue:43723|issue:44554", + "issue:40444|issue:43526", "issue:43335|issue:44112", + "issue:43723|issue:44554", "issue:34689|issue:35141", "issue:44857|issue:45072", "issue:43606|issue:44857", - "issue:45372|issue:45406", "issue:44977|issue:45563", + "issue:45372|issue:45406", "issue:43066|issue:44462", "issue:43873|issue:44493", "issue:43873|issue:45229", @@ -6888,248 +6924,252 @@ "issue:41720|issue:45229", "issue:45245|issue:45563", "issue:32090|issue:33357", - "issue:43976|issue:45440", - "issue:37428|issue:44987", - "issue:45372|issue:45459" + "issue:37428|issue:44987" ], - "prepared_review_unit_hash": "3cf32d8dde04e3ec2e05f226fc6c409f661619fcee9df6791cb37c21a39f459c", + "prepared_review_unit_hash": "a8b3e0fece76bda5f1f91c56d74a3d231e8cc75193a7ebf373ef5088e0b01235", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 38348, - "estimated_input_tokens": 9587, - "estimated_eval_tokens": 19430 + "soft_pair_count": 36, + "serialized_chars": 38743, + "estimated_input_tokens": 9686, + "estimated_eval_tokens": 19628 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous; none of the soft-similarity pairs look like true duplicates or the same fix. The closest hub is issue 43335, but it is still only a local config bug among many unrelated model/tokenizer/runtime problems.", - "confidence": 0.95, - "canonical_issue_reason": "issue:43335 is the closest thing to a hub in this set: it is a concrete, self-contained bug with multiple weak similarity links, even though the overall cluster is not a single duplicate family.", + "summary": "This cluster is mostly a bag of unrelated Transformer issues across different models and subsystems. I found no convincing duplicate pairs to merge; the only reasonably strong standalone triage target is issue 45229 because it is open, high-activity, and has a concrete reproduction with broad user impact.", + "confidence": 0.84, + "canonical_issue_reason": "No clear duplicate cluster emerged; if one issue must represent the set, issue 45229 is the strongest standalone bug report thanks to detailed discussion and impact.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43335 is the best representative anchor because it describes a specific code-path bug with a plausible single fix, whereas the rest of the cluster is a grab bag of unrelated regressions.", + "best_issue_reason": "Issue 45229 is the most actionable and representative issue in the set: open, heavily discussed, and framed around a concrete runtime failure.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43404", - "right": "issue:45200", + "left": "issue:44855", + "right": "issue:45083", "accept": false, - "reason": "Different models and failures: Mistral3 lm_head tying vs Gemma 4 multimodal token-type handling." + "reason": "Different failures: Python 3.13 import/parsing error in DebertaV2 vs a Qwen3 omni feature-length helper bug." }, { - "left": "issue:29127", - "right": "issue:32090", + "left": "issue:45200", + "right": "issue:45381", + "accept": false, + "reason": "Both are multimodal, but one is about missing mm_token_type_ids defaults and the other is a Qwen2.5-VL video vision_position_ids bug." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "Doc/example failures for SigLIP2 are not the same as a generic position-id key warning across many models." + }, + { + "left": "issue:43404", + "right": "issue:45200", "accept": false, - "reason": "Unrelated error classes: layoutlmv3 box-information message vs Trainer NoneType GPU broadcast crash." + "reason": "Mistral3 lm_head tying and Gemma 4 mm_token_type_ids are unrelated code paths." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "Different subsystems: SwitchTransformers sparse-layer config bug vs BigBirdTokenizer special-token registration." + "reason": "SwitchTransformers sparse-layer config bug is unrelated to BigBirdTokenizer special-token registration." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "An error-message clarity issue in LayoutLMv3 is not the same as a Trainer broadcast TypeError." }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Both are v4/v5 output regressions, but on different tokenizer/model paths and different root causes." + "reason": "Both involve v5 behavior changes, but one is Pegasus output drift and the other is Deepseek tokenizer correctness; different model/code paths." }, { "left": "issue:44977", "right": "issue:45229", "accept": false, - "reason": "Different issues: flash-attention generation failure vs multi-GPU CUDA OOM." + "reason": "Qwen3.5 flash-attention generation failure and Gemma4 multi-GPU CUDA OOM are different symptoms with different root causes." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "Different symptoms and code paths: zero probabilities in SigLIP2 vs unexpected position-id key warnings." + "reason": "Unexpected zero probabilities in SigLIP2 is not the same bug as widespread unexpected position-id keys." }, { "left": "issue:44292", "right": "issue:45520", "accept": false, - "reason": "Qwen-NVFP4 runtime failure is unrelated to the Python 3.13 flash_attn import-key error." + "reason": "Qwen-3-8B-NVFP4 runtime failure and a Python 3.13 flash_attn import KeyError are unrelated." }, { "left": "issue:43335", "right": "issue:43504", "accept": false, - "reason": "SwitchTransformers config bug and BEiT legacy-field loading bug are unrelated." + "reason": "Sparse-layer creation in SwitchTransformers and BEiT legacy-field loading are different bugs." }, { "left": "issue:43335", "right": "issue:45083", "accept": false, - "reason": "Different model families and code paths; not the same underlying bug." + "reason": "A SwitchTransformers config edge case does not match a Qwen3 Omni feature-length helper regression." }, { "left": "issue:43550", "right": "issue:44112", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is unrelated to the GraniteSpeech stale-test device override issue." + "reason": "torch.compile/SDPA failure in Bamba is unrelated to a stale GraniteSpeech CI test override." }, { "left": "issue:43335", "right": "issue:45072", "accept": false, - "reason": "Sparse-layer creation bug vs bfloat16 dtype mismatch in inference are different problems." + "reason": "SwitchTransformers config logic and bfloat16 dtype mismatches in inference are not the same underlying issue." }, { "left": "issue:44568", "right": "issue:45491", "accept": false, - "reason": "Tokenizer special-token behavior does not match the Gemma3 NaN attention/embedding failure." + "reason": "Tokenizer special-token handling for mdeberta-v3-base is unrelated to Gemma3 NaN embeddings from sliding-window attention." }, { "left": "issue:43720", "right": "issue:44112", "accept": false, - "reason": "Packed-weight loading bug and a CI test/device-override issue are not the same change." - }, - { - "left": "issue:40444", - "right": "issue:43526", - "accept": false, - "reason": "Qwen2.5-VL iterable dataset handling is unrelated to BEiT fast processor label reduction." + "reason": "BitNet packed-weight loading and a stale CI test are completely different concerns." }, { "left": "issue:44448", "right": "issue:44977", "accept": false, - "reason": "Different regressions: Pegasus v4/v5 output drift vs Qwen3.5 flash-attention generation failure." + "reason": "Pegasus v4/v5 output differences and Qwen3.5 flash-attention generation issues are distinct model regressions." }, { - "left": "issue:43723", - "right": "issue:44554", + "left": "issue:40444", + "right": "issue:43526", "accept": false, - "reason": "Tokenizer loading bug and an MPS attention correctness issue are different code paths." + "reason": "Qwen2.5-VL iterable-dataset multi-image finetuning failure is unrelated to BEiT fast processor label reduction." }, { "left": "issue:43335", "right": "issue:44112", "accept": false, - "reason": "No shared bug; one is SwitchTransformers config creation, the other is a stale CI device-override test." + "reason": "SwitchTransformers sparse-layer config and GraniteSpeech device-override CI failure do not share a root bug." + }, + { + "left": "issue:43723", + "right": "issue:44554", + "accept": false, + "reason": "Tokenizer loading in v5 and an MPS attention correctness issue are different subsystems and failures." }, { "left": "issue:34689", "right": "issue:35141", "accept": false, - "reason": "Llama 3.2 Vision model loading regression is unrelated to token-embedding reinitialization on resize." + "reason": "Llama 3.2 Vision model loading regression and embedding retie after resize are not the same bug." }, { "left": "issue:44857", "right": "issue:45072", "accept": false, - "reason": "Different dtype/device problems in different models and execution modes." + "reason": "Float16 AMP crash in LwDetrImageLoss is unrelated to bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:43606", "right": "issue:44857", "accept": false, - "reason": "CPU offload device mismatch is not the same as float16 AMP crash in LwDetrImageLoss." + "reason": "CPU offload device mismatch and an AMP/CUDA float16 crash are separate runtime issues." }, { - "left": "issue:45372", - "right": "issue:45406", + "left": "issue:44977", + "right": "issue:45563", "accept": false, - "reason": "Both involve Gemma 4, but one is a missing mistral_common import and the other is a missing _tokenizer attribute; different root causes." + "reason": "Flash-attention generation failure and a stale warning about num_return_sequences are not the same problem." }, { - "left": "issue:44977", - "right": "issue:45563", + "left": "issue:45372", + "right": "issue:45406", "accept": false, - "reason": "Generation instability with flash-attention is unrelated to the stale warning about num_return_sequences." + "reason": "Both involve Gemma4, but one is a mistral_common import error during processor loading and the other is a missing _tokenizer attribute in serve." }, { "left": "issue:43066", "right": "issue:44462", "accept": false, - "reason": "Wrong tokenizer decoder type in v5 is different from AutoTokenizer ignoring tokenizer.json." + "reason": "Wrong tokenizer decoder type and ignoring tokenizer.json are related broadly to tokenizers, but they target different behaviors and fixes." }, { "left": "issue:43873", "right": "issue:44493", "accept": false, - "reason": "Quantization/offloading behavior and position-id key warnings are unrelated." + "reason": "Quantization/offloading behavior and unexpected position-id keys are unrelated." }, { "left": "issue:43873", "right": "issue:45229", "accept": false, - "reason": "Offloading/quantization issue is not the same as Gemma4 multi-GPU OOM." + "reason": "Offloading with quantization and Gemma4 multi-GPU CUDA OOM are different resource-management failures." }, { "left": "issue:43278", "right": "issue:43329", "accept": false, - "reason": "Embedding dtype drift and multimodal token-count helper bugs are different problems." + "reason": "Embedding dtype drift between train/eval is unrelated to the multimodal token-count video-branch bug." }, { "left": "issue:43701", "right": "issue:45083", "accept": false, - "reason": "Checkpoint key mismatch is unrelated to qwen3_omni_moe feature-length calculation." + "reason": "resume_from_checkpoint key mismatch does not match the Qwen3 Omni feature-length helper issue." }, { "left": "issue:29942", "right": "issue:35141", "accept": false, - "reason": "Flash Attention 2 test failures are not the same as embedding reinitialization on resize." + "reason": "Flash Attention 2 test failures are unrelated to embedding reinitialization after token resize." }, { "left": "issue:43550", "right": "issue:45083", "accept": false, - "reason": "torch.compile/SDPA failure is unrelated to qwen3_omni_moe output-length helper behavior." + "reason": "Bamba torch.compile/SDPA failure is not the same as a Qwen3 Omni helper bug." }, { "left": "issue:41720", "right": "issue:45229", "accept": false, - "reason": "Different symptoms and likely different layers: auto device mapping cudaErrorAssert vs inference OOM." + "reason": "Qwen3 auto device-mapping assert and Gemma4 multi-GPU OOM are different model/runtime failures." }, { "left": "issue:45245", "right": "issue:45563", "accept": false, - "reason": "Category-count runtime limit and stale warning emission are unrelated." + "reason": "Category-count limit error and a stale paged-generate warning are unrelated." }, { "left": "issue:32090", "right": "issue:33357", "accept": false, - "reason": "Trainer NoneType broadcast crash is unrelated to MacOS bus error on community CLIP." - }, - { - "left": "issue:43976", - "right": "issue:45440", - "accept": false, - "reason": "Python version incompatibility is unrelated to DeepseekV3MoE remote-vs-native divergence." + "reason": "Trainer NoneType broadcast failure and a MacOS CLIP bus error are different issues." }, { "left": "issue:37428", "right": "issue:44987", "accept": false, - "reason": "Missing flash_attention_utils import and a model-loading failure for physical-intelligence/fast are not the same bug." - }, - { - "left": "issue:45372", - "right": "issue:45459", - "accept": false, - "reason": "Gemma 4 processor import failure and tokenizer errors hidden by protobuf handling are different issues." + "reason": "Flash-attention import-name failure in Python 3.13 and a transformers>=5.1.0 model-loading failure are different root causes." } ] }, @@ -7170,11 +7210,11 @@ "issue:43381", "issue:43421", "issue:43504", - "issue:43526", "issue:43653", "issue:43701", "issue:43756", "issue:43824", + "issue:43976", "issue:43994", "issue:44186", "issue:44206", @@ -7184,7 +7224,6 @@ "issue:44442", "issue:44448", "issue:44451", - "issue:44464", "issue:44466", "issue:44479", "issue:44610", @@ -7199,15 +7238,18 @@ "issue:45372", "issue:45412", "issue:45440", + "issue:45459", "issue:45491", "issue:45520" ], "soft_pairs": [ + "issue:43976|issue:45440", + "issue:45372|issue:45459", "issue:44448|issue:44466", "issue:44451|issue:45520", - "issue:43065|issue:43994", - "issue:43065|issue:45412", "issue:25251|issue:32090", + "issue:43065|issue:45412", + "issue:43065|issue:43994", "issue:44186|issue:45198", "issue:43278|issue:43381", "issue:43653|issue:44743", @@ -7221,8 +7263,8 @@ "issue:44206|issue:44479", "issue:43756|issue:45440", "issue:43504|issue:44857", - "issue:44610|issue:45412", "issue:29942|issue:30064", + "issue:44610|issue:45412", "issue:33357|issue:39692", "issue:32090|issue:33290", "issue:45103|issue:45372", @@ -7230,218 +7272,216 @@ "issue:45072|issue:45229", "issue:43824|issue:44279", "issue:43381|issue:43701", - "issue:43653|issue:44442", - "issue:43526|issue:44186", - "issue:44464|issue:45491" + "issue:43653|issue:44442" ], - "prepared_review_unit_hash": "3ada753b45e63b7c7fefaa5f00f41c4dabc037cc82ee0f0623c6f3b464341cbb", + "prepared_review_unit_hash": "7bedd4ceaa3381622f7c6ec11657994b95725c2893c40df267074164ef6bdfaf", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36689, - "estimated_input_tokens": 9173, - "estimated_eval_tokens": 18602 + "serialized_chars": 36666, + "estimated_input_tokens": 9167, + "estimated_eval_tokens": 18590 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the issues span unrelated tokenizer, vision, training, quantization, import, and performance regressions. I do not see a single underlying bug or change that ties the soft-edge pairs together, so all candidate duplicate links are rejected.", - "confidence": 0.21, - "canonical_issue_reason": null, + "summary": "This is a heterogeneous similarity bucket, not a duplicate cluster. The suggested pairs share broad library/model keywords but describe different bugs, regressions, or features, so none should be merged as the same underlying issue.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items are unrelated enough that none is a defensible duplicate anchor.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43994 is a clear, self-contained user-facing bug report with a specific model, reproducible bad output, and a concrete failure mode; it is the strongest standalone representative in this mixed set.", + "best_issue_reason": "No single issue is a good global representative for this cluster; the issues span distinct subsystems and failure modes.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python-version compatibility issue vs DeepSeekV3 implementation divergence; different code paths and failures." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma 4 processor import failure vs tokenizer error masking when protobuf is missing; not the same bug." + }, { "left": "issue:44448", "right": "issue:44466", "accept": false, - "reason": "Different regressions: one is pipeline top_k output shape, the other is device-dependent lm_head serialization." + "reason": "Different problems: Pegasus output regression vs inconsistent lm_head serialization." }, { "left": "issue:44451", "right": "issue:45520", "accept": false, - "reason": "Unrelated failures: model loading for ScandiBERT vs a Python 3.13 flash_attn import KeyError." + "reason": "Model loading failure for a specific checkpoint vs flash_attn import KeyError on Python 3.13." }, { - "left": "issue:43065", - "right": "issue:43994", + "left": "issue:25251", + "right": "issue:32090", "accept": false, - "reason": "Different model areas and symptoms; a dummy Conv2d in SAM3 pixel decoding is not the same as SigLIP2 producing nonsense outputs." + "reason": "Pipeline top_k nesting bug is unrelated to Trainer _gpu_broadcast_one NoneType error." }, { "left": "issue:43065", "right": "issue:45412", "accept": false, - "reason": "A pixel-decoder placeholder layer and an RT-DETR memory leak are unrelated bugs." + "reason": "Dummy Conv2d in Sam3PixelDecoder vs RT-DETR memory not being released; different defects." }, { - "left": "issue:25251", - "right": "issue:32090", + "left": "issue:43065", + "right": "issue:43994", "accept": false, - "reason": "Pipeline top_k nesting behavior is unrelated to Trainer GPU broadcast NoneType errors." + "reason": "SAM3 pixel decoder placeholder layer vs SigLIP2 AutoModel/pipeline misbehavior." }, { "left": "issue:44186", "right": "issue:45198", "accept": false, - "reason": "Different tokenizers and different failure modes: LayoutLMv2 NER padding crashes vs Wav2Vec2 save/tokenization failures." + "reason": "LayoutLMv2 tokenizer crash on NER/padding vs Wav2Vec2 save_pretrained/tokenization failure." }, { "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "One is dtype drift between train/eval, the other is a checkpointing restriction in eval mode; not the same bug." + "reason": "Embedding dtype drift in evaluation vs gradient checkpointing being unsupported in eval mode." }, { "left": "issue:43653", "right": "issue:44743", "accept": false, - "reason": "Tokenizer special-token registration and Qwen recurrent-state reset are unrelated." + "reason": "BigBird special-token registration bug vs Qwen3.5 recurrent-state reset bug." }, { "left": "issue:44265", "right": "issue:45198", "accept": false, - "reason": "torch.export exportability via torch_compilable_check is a different code path from Wav2Vec2 save_pretrained/tokenization." + "reason": "torch.export/torch_compilable_check failure vs Wav2Vec2 save/tokenization failure." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "MLukeTokenizer task AttributeError is unrelated to the Wav2Vec2 serialization/tokenization issue." + "reason": "MLukeTokenizer AttributeError vs Wav2Vec2 save/tokenization failure; unrelated components." }, { "left": "issue:43257", "right": "issue:43278", "accept": false, - "reason": "Qwen3 MoE weight conversion under accelerate/deepspeed is unrelated to embedding dtype changes in eval." + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed vs embedding dtype mismatch during eval." }, { "left": "issue:44871", "right": "issue:44977", "accept": false, - "reason": "Gemma EOS-token config mismatch and Qwen3.5 flash-attention generation failure are different problems." + "reason": "Gemma eos_token_id config inconsistency vs Qwen3.5 flash-attention generation regression." }, { "left": "issue:43421", "right": "issue:45491", "accept": false, - "reason": "Runtime post-processor updates for special tokens are unrelated to Gemma3 NaN embeddings from sliding-window attention." + "reason": "Runtime post-processor update bug in TokenizersBackend vs NaN embeddings from Gemma3 sliding-window attention." }, { "left": "issue:43994", "right": "issue:45412", "accept": false, - "reason": "SigLIP2 inference nonsense and RT-DETR memory release behavior are different bugs." + "reason": "SigLIP2 inference/output issue vs RT-DETR memory leak; no shared failing code path." }, { "left": "issue:42915", "right": "issue:42994", "accept": false, - "reason": "Both involve quantization, but one is FP8 model loading and the other is saving quantized models; not the same defect." + "reason": "FineGrainedFP8Config failure vs quantized model saving failure are different quantization problems." }, { "left": "issue:44206", "right": "issue:44479", "accept": false, - "reason": "Unsupported feature-extractor args for LasrFeatureExtractor are unrelated to Qwen VL video input regressions." + "reason": "Unsupported center argument in LasrFeatureExtractor vs Qwen VL video-input regression." }, { "left": "issue:43756", "right": "issue:45440", "accept": false, - "reason": "Smollm3 RoPE-layer mismatch and DeepseekV3 implementation divergence are different model-architecture issues." + "reason": "Smollm3 RoPE-layer mismatch vs DeepSeekV3 implementation divergence; different architectures and symptoms." }, { "left": "issue:43504", "right": "issue:44857", "accept": false, - "reason": "Legacy preset loading in BEiT segmentation and float16 CUDA loss crashes are unrelated." + "reason": "Legacy field prevents BEiT preset loading vs LwDetrImageLoss AMP/CUDA crash." }, { - "left": "issue:44610", - "right": "issue:45412", + "left": "issue:29942", + "right": "issue:30064", "accept": false, - "reason": "Processor image-size mismatch for OmDet-Turbo is unrelated to RT-DETR memory cleanup." + "reason": "Flash Attention 2 test failures vs void segmentation-map processing bug." }, { - "left": "issue:29942", - "right": "issue:30064", + "left": "issue:44610", + "right": "issue:45412", "accept": false, - "reason": "Flash Attention 2 test failures and void segmentation-map processing failures are different failures in different subsystems." + "reason": "Processor/model input-size mismatch vs RT-DETR garbage-collection memory issue." }, { "left": "issue:33357", "right": "issue:39692", "accept": false, - "reason": "MacOS CLIP bus error and SigLIP2 documentation/example errors are not the same bug." + "reason": "MacOS bus error on CLIP model vs SigLIP2 documentation example errors; distinct models and failures." }, { "left": "issue:32090", "right": "issue:33290", "accept": false, - "reason": "NoneType broadcast failure in Trainer is unrelated to Adafactor + DeepSpeed OOM." + "reason": "Trainer broadcast NoneType error vs Adafactor+DeepSpeed OOM; unrelated causes." }, { "left": "issue:45103", "right": "issue:45372", "accept": false, - "reason": "auto_docstring annotation parsing and Gemma4 processor import errors involve different modules and causes." + "reason": "auto_docstring annotations crash vs Gemma 4 processor import dependency issue." }, { "left": "issue:29127", "right": "issue:34567", "accept": false, - "reason": "LayoutLMv3 error messaging and TrainerState token-count tracking are unrelated." + "reason": "LayoutLMv3 error-message clarity vs TrainerState token-count tracking bug." }, { "left": "issue:45072", "right": "issue:45229", "accept": false, - "reason": "bfloat16 dtype mismatch in CI is unrelated to Gemma4 31B multi-GPU OOM." + "reason": "bfloat16 dtype mismatches in inference vs Gemma4 multi-GPU CUDA OOM." }, { "left": "issue:43824", "right": "issue:44279", "accept": false, - "reason": "Missing Qwen2_5_VL import and a generic dependency issue are not the same concrete bug." + "reason": "Missing Qwen2_5_VL import vs generic transformers dependency issue; not the same bug." }, { "left": "issue:43381", "right": "issue:43701", "accept": false, - "reason": "Eval-mode gradient checkpointing restriction and resume-from-checkpoint key mismatch are different training issues." + "reason": "Gradient checkpointing in eval mode vs resume_from_checkpoint key mismatch." }, { "left": "issue:43653", "right": "issue:44442", "accept": false, - "reason": "BigBird tokenizer special-token registration and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are unrelated." - }, - { - "left": "issue:43526", - "right": "issue:44186", - "accept": false, - "reason": "BEiT fast processor label reduction and LayoutLMv2 tokenizer crashes are different code paths." - }, - { - "left": "issue:44464", - "right": "issue:45491", - "accept": false, - "reason": "Chunked generation with compiled forward and Gemma3 NaN embeddings are distinct generation/attention bugs." + "reason": "BigBird special-token registration bug vs AutoTokenizer loading failure for FastSpeech2ConformerTokenizer." } ] }, @@ -7467,10 +7507,8 @@ "nodes": [ "issue:29127", "issue:29942", - "issue:33290", "issue:33357", "issue:33666", - "issue:34567", "issue:35141", "issue:36010", "issue:39692", @@ -7485,7 +7523,6 @@ "issue:43531", "issue:43575", "issue:43646", - "issue:43673", "issue:43701", "issue:43723", "issue:43761", @@ -7493,10 +7530,12 @@ "issue:43994", "issue:44077", "issue:44079", + "issue:44186", "issue:44188", "issue:44206", "issue:44387", "issue:44442", + "issue:44464", "issue:44479", "issue:44610", "issue:44743", @@ -7515,20 +7554,22 @@ "issue:45491" ], "soft_pairs": [ + "issue:44464|issue:45491", + "issue:43526|issue:44186", "issue:43334|issue:43531", "issue:43262|issue:44206", - "issue:44977|issue:45245", "issue:29942|issue:36010", + "issue:44977|issue:45245", "issue:29127|issue:33357", "issue:29127|issue:35141", "issue:43761|issue:44079", - "issue:42915|issue:43931", "issue:44442|issue:45072", + "issue:42915|issue:43931", "issue:43232|issue:43388", "issue:43526|issue:44857", - "issue:44479|issue:45290", "issue:43526|issue:44610", "issue:44743|issue:44811", + "issue:44479|issue:45290", "issue:43723|issue:45479", "issue:45003|issue:45440", "issue:43575|issue:43701", @@ -7536,217 +7577,215 @@ "issue:43646|issue:44077", "issue:43761|issue:44188", "issue:43575|issue:44387", - "issue:43994|issue:44206", "issue:44964|issue:45440", - "issue:43316|issue:43525", "issue:29942|issue:33666", + "issue:43994|issue:44206", + "issue:43316|issue:43525", "issue:39692|issue:45491", - "issue:44442|issue:44898", - "issue:33290|issue:34567", - "issue:43673|issue:44743" + "issue:44442|issue:44898" ], - "prepared_review_unit_hash": "ea6bdb103755c5b90b3625240cb68040878cfa45ae9c60692bf21c64e55846d2", + "prepared_review_unit_hash": "d836530ed7e88d8f737171749d62132d7285e72f7cd53635d34fa71e0b2585ce", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 36434, - "estimated_input_tokens": 9109, - "estimated_eval_tokens": 18474 + "serialized_chars": 35919, + "estimated_input_tokens": 8980, + "estimated_eval_tokens": 18216 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a true duplicate set: the items span many unrelated model-, training-, processor-, and generation-specific bugs. The soft links are mostly lexical similarity rather than the same underlying defect, so all candidate merges should be rejected.", - "confidence": 0.37, - "canonical_issue_reason": "If a single representative is needed, issue:45003 is the broadest/open-ended bug in the set, but it is still only a weak proxy because the cluster is heterogeneous.", + "summary": "The set is highly heterogeneous: most items are unrelated regressions across different models, processors, and APIs. I found no soft pair that clearly represents the same underlying bug/change, so I would not merge any of the candidate edges.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 43646 is the broadest version-regression item in the set and the best single umbrella anchor if one must be chosen, but the overall cluster is too mixed for a true canonical duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45003 is the best available umbrella representative due to its broad modeling_utils scope, but it does not actually match most of the other reports.", + "best_issue_reason": "43646 is the most general, high-level issue title among otherwise unrelated reports, so it is the safest representative issue for triage, though not a real duplicate hub.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44464", + "right": "issue:45491", + "accept": false, + "reason": "Both involve generation/attention behavior, but one is chunked generation with compiled forward and the other is Gemma3 NaN embeddings on mixed-length batches; different code paths and failures." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Both are preprocessing/label-handling bugs, but they affect different components and symptoms (BeitImageProcessorFast label reduction vs LayoutLMv2Tokenizer NER/padding crashes)." + }, { "left": "issue:43334", "right": "issue:43531", "accept": false, - "reason": "Different problems: missing pad_token_id on Qwen3-VL vs a sliding-window issue on Qwen3-MoE." + "reason": "Both are Qwen3-family issues, but one is a missing `pad_token_id` on load and the other is a `sliding_window` behavior bug in Qwen3-MoE; not the same defect." }, { "left": "issue:43262", "right": "issue:44206", "accept": false, - "reason": "Audio chat-template sampling-rate default bug vs a feature extractor passing an unsupported center arg." + "reason": "Both touch audio processing, but one is a chat-template sampling-rate default and the other is an unsupported `center` argument in a feature extractor; distinct bugs." }, { - "left": "issue:44977", - "right": "issue:45245", + "left": "issue:29942", + "right": "issue:36010", "accept": false, - "reason": "Both mention failures, but one is a flash-attention generation regression and the other is an unrelated category-count RuntimeError." + "reason": "FlashAttention test failures and a `GenerationMixin` import error are unrelated failure modes with no shared concrete code path." }, { - "left": "issue:29942", - "right": "issue:36010", + "left": "issue:44977", + "right": "issue:45245", "accept": false, - "reason": "Flash Attention 2 test failures are unrelated to the GenerationMixin import error." + "reason": "Qwen3.5 flash-attention generation failure and the 2^24 category limit error are entirely different problems." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "LayoutLMv3 error messaging issue vs a CLIP bus error on macOS; different models and failure modes." + "reason": "LayoutLMv3 error messaging and a MacOS bus error in a CLIP community model are unrelated." }, { "left": "issue:29127", "right": "issue:35141", "accept": false, - "reason": "Clarifying a LayoutLMv3 boxes error is unrelated to output embedding reinitialization after resize." + "reason": "Different subsystems: LayoutLMv3 box-info validation vs embedding resize/post-init reinitialization." }, { "left": "issue:43761", "right": "issue:44079", "accept": false, - "reason": "CLIPVisionModel hidden_states regression is distinct from ModelOutput key assignment behavior." + "reason": "One is a CLIPVisionModel hidden-states regression; the other is a generic `ModelOutput` key-assignment issue. They may be loosely related, but not the same concrete bug." }, { - "left": "issue:42915", - "right": "issue:43931", + "left": "issue:44442", + "right": "issue:45072", "accept": false, - "reason": "FineGrainedFP8Config failure on Qwen3Moe is not the same as Qwen3-VL weight-shape mismatch at load time." + "reason": "AutoTokenizer loading failure and bfloat16 dtype mismatches are different classes of bugs." }, { - "left": "issue:44442", - "right": "issue:45072", + "left": "issue:42915", + "right": "issue:43931", "accept": false, - "reason": "Tokenizer loading failure and dtype mismatch in bfloat16 inference are unrelated issues." + "reason": "Both mention Qwen3 models, but one is FP8 quantization failure and the other is a weight-shape mismatch on load; no clear shared code path." }, { "left": "issue:43232", "right": "issue:43388", "accept": false, - "reason": "Generation kwargs synchronization bug is unrelated to gather_for_metrics dropping tuple labels." + "reason": "Generation kwargs syncing and metric-gathering label truncation are unrelated training/runtime paths." }, { "left": "issue:43526", "right": "issue:44857", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug is not the same as LwDetrImageLoss AMP/CUDA crash." - }, - { - "left": "issue:44479", - "right": "issue:45290", - "accept": false, - "reason": "Video input regression for Qwen VL models is distinct from chat-template crashes on tool-call assistant messages." + "reason": "A label reduction bug in BeitImageProcessorFast is not the same as an AMP/CUDA crash in LwDetrImageLoss." }, { "left": "issue:43526", "right": "issue:44610", "accept": false, - "reason": "Both are processor-related, but one is label reduction and the other is OmDet-Turbo image size mismatch." + "reason": "Both concern image preprocessing, but one is label reduction and the other is incorrect output resolution; different underlying bugs." }, { "left": "issue:44743", "right": "issue:44811", "accept": false, - "reason": "Recurrent state reset in modular_qwen3_5.py does not match Whisper batch_decode skip_special_tokens handling." + "reason": "Qwen3.5 recurrent-state reset and Whisper `batch_decode` skipping special tokens are unrelated issues." + }, + { + "left": "issue:44479", + "right": "issue:45290", + "accept": false, + "reason": "One is a video-input regression across Qwen VL models; the other is a chat-template crash on assistant tool-call messages. Different features and code paths." }, { "left": "issue:43723", "right": "issue:45479", "accept": false, - "reason": "Tokenizer loading regression in v5 is unrelated to single-label classification producing zero loss." + "reason": "Tokenizer loading in v5 and single-label classification zero-loss are unrelated." }, { "left": "issue:45003", "right": "issue:45440", "accept": false, - "reason": "Unsafe sys.modules access in modeling_utils is unrelated to DeepSeekV3MoE divergence from remote implementation." + "reason": "Unsafe `sys.modules` access in modeling_utils and a DeepSeekV3MoE divergence are separate implementation areas." }, { "left": "issue:43575", "right": "issue:43701", "accept": false, - "reason": "Tensor-parallel OOM on Qwen2-57B is not the same as resume_from_checkpoint key mismatch." + "reason": "Tensor-parallel OOM on model load and resume-from-checkpoint key mismatch are not the same issue." }, { "left": "issue:43931", "right": "issue:45216", "accept": false, - "reason": "Qwen3-VL load-time weight mismatch is different from Qwen3.5 save_pretrained checkpoint corruption." + "reason": "Both are Qwen3.5/Qwen3-VL loading/saving issues, but one is a checkpoint shape mismatch and the other is an incorrect `save_pretrained` checkpoint; related family, different bug." }, { "left": "issue:43646", "right": "issue:44077", "accept": false, - "reason": "Custom model initialization breakage in Transformers 5.0 is not the same as patchtsmixer's post_init policy change." + "reason": "Generic custom-model init breakage and the `patchtsmixer` post_init policy issue do not share the same failure mechanism." }, { "left": "issue:43761", "right": "issue:44188", "accept": false, - "reason": "CLIPVisionModel hidden_states regression is unrelated to divergent attention kernels under torch.compile." + "reason": "A CLIPVisionModel output regression and a torch.compile attention-kernel divergence are different code paths." }, { "left": "issue:43575", "right": "issue:44387", "accept": false, - "reason": "Model load OOM with tensor parallelism is distinct from increased CUDA reserved memory under int4 quantization." - }, - { - "left": "issue:43994", - "right": "issue:44206", - "accept": false, - "reason": "SigLIP2 inference/pipeline issue is not the same as the unsupported center arg crash in LasrFeatureExtractor." + "reason": "OOM from tensor-parallel loading and increased reserved memory under int4 quantization are both memory issues, but not the same root cause." }, { "left": "issue:44964", "right": "issue:45440", "accept": false, - "reason": "Phi-4 multimodal load failure is unrelated to DeepSeekV3MoE behavioral divergence." - }, - { - "left": "issue:43316", - "right": "issue:43525", - "accept": false, - "reason": "Gemma3TextConfig API discrepancy and Llama4Config missing pad_token_id are similar in shape but not the same bug." + "reason": "Phi-4 multimodal load failure and DeepSeekV3MoE divergence are unrelated model-specific issues." }, { "left": "issue:29942", "right": "issue:33666", "accept": false, - "reason": "Flash Attention 2 test failures are unrelated to Qwen2-VL multi-GPU training." + "reason": "Flash Attention 2 test failures and Qwen2-VL multi-GPU training are different problem domains." }, { - "left": "issue:39692", - "right": "issue:45491", + "left": "issue:43994", + "right": "issue:44206", "accept": false, - "reason": "SigLIP2 documentation example errors are not the same as Gemma3 NaN embeddings from mixed-length batching." + "reason": "SigLIP2 model/pipeline nonsense results and an audio feature extractor argument crash are unrelated." }, { - "left": "issue:44442", - "right": "issue:44898", + "left": "issue:43316", + "right": "issue:43525", "accept": false, - "reason": "Tokenizer loading failure and Perceiver non-default resolution failure are separate model/processor bugs." + "reason": "Both mention config API discrepancies, but they affect different configs and different missing attributes; too broad to be a duplicate." }, { - "left": "issue:33290", - "right": "issue:34567", + "left": "issue:39692", + "right": "issue:45491", "accept": false, - "reason": "Adafactor OOM in deepspeed is unrelated to TrainerState num_input_tokens_seen not updating." + "reason": "SigLIP2 doc-example problems and Gemma3 NaN embeddings are unrelated." }, { - "left": "issue:43673", - "right": "issue:44743", + "left": "issue:44442", + "right": "issue:44898", "accept": false, - "reason": "Missing GenerationMixin cache in v5.0.0 is not the same as recurrent states resetting in Qwen3.5 modular code." + "reason": "AutoTokenizer loading failure and Perceiver non-default-resolution failure are different bugs in different components." } ] }, @@ -7775,13 +7814,13 @@ "issue:25251", "issue:28282", "issue:33290", + "issue:34567", "issue:34689", "issue:35141", "issue:36010", "issue:36331", "issue:39692", "issue:41628", - "issue:41950", "issue:42175", "issue:43065", "issue:43316", @@ -7789,13 +7828,12 @@ "issue:43388", "issue:43404", "issue:43441", - "issue:43475", "issue:43493", "issue:43540", "issue:43611", "issue:43643", + "issue:43673", "issue:43688", - "issue:43716", "issue:43720", "issue:43749", "issue:43827", @@ -7811,6 +7849,7 @@ "issue:44496", "issue:44514", "issue:44617", + "issue:44743", "issue:44792", "issue:44855", "issue:44871", @@ -7820,21 +7859,23 @@ "issue:45372" ], "soft_pairs": [ + "issue:33290|issue:34567", + "issue:43673|issue:44743", "issue:34689|issue:43611", "issue:43381|issue:43873", "issue:43493|issue:43827", "issue:33290|issue:35141", "issue:33290|issue:34689", "issue:44077|issue:44479", - "issue:41628|issue:45372", "issue:43720|issue:44442", + "issue:41628|issue:45372", "issue:43749|issue:43931", "issue:43643|issue:44913", "issue:43873|issue:44871", "issue:39692|issue:44206", "issue:44190|issue:44496", - "issue:33290|issue:36331", "issue:43381|issue:43856", + "issue:33290|issue:36331", "issue:34689|issue:36010", "issue:44855|issue:45137", "issue:43540|issue:44514", @@ -7845,206 +7886,204 @@ "issue:16998|issue:22355", "issue:43404|issue:43441", "issue:25251|issue:28282", - "issue:43065|issue:44617", - "issue:41950|issue:43475", - "issue:43316|issue:43716" + "issue:43065|issue:44617" ], - "prepared_review_unit_hash": "e48eaaa041a75a58a5d50488235c6f805e4af9034ad520bbc504570dad2e851a", + "prepared_review_unit_hash": "74dc5a548a9309b5d734550bac2e433f131f9fe0ecd938b5b5e7bafaf8d8887e", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36045, - "estimated_input_tokens": 9012, - "estimated_eval_tokens": 18280 + "serialized_chars": 36072, + "estimated_input_tokens": 9018, + "estimated_eval_tokens": 18292 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is heterogeneous; none of the proposed soft edges look like true duplicates. The strongest standalone issue to anchor the cluster is #43749, but it does not match the others closely enough to merge.", - "confidence": 0.95, - "canonical_issue_reason": "If a single issue must represent the set, #43749 is the most central by discussion/inbound refs and describes a concrete loader bug; the rest are largely unrelated.", + "summary": "These items are mostly unrelated Transformers issues with overlapping vocabulary. None of the soft pairs look like true duplicates or mergeable equivalents of the same underlying bug/change.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits this cluster: the items span unrelated imports, configs, training bugs, model-loading regressions, and docs problems, so there is no shared root cause to dedupe around.", "canonical_pr_reason": null, - "best_issue_reason": "#43749 is the most actionable and highest-signal issue here, but it is only a representative, not a duplicate hub.", + "best_issue_reason": "No issue is a good global representative; the similarity here is mostly lexical, not causal, and the problems target different code paths and models.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:33290", + "right": "issue:34567", + "accept": false, + "reason": "Optimizer OOM in DeepSpeed and TrainerState token-count update are different training subsystems and bugs." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Missing GenerationMixin cache and Qwen3.5 recurrent-state reset are different generation/cache issues." + }, { "left": "issue:34689", "right": "issue:43611", "accept": false, - "reason": "Both are load-time regressions, but they hit different model paths and failures." + "reason": "Both are loading regressions, but they affect different mechanisms and models: Llama 3.2 vision loading vs base_model_prefix handling." }, { "left": "issue:43381", "right": "issue:43873", "accept": false, - "reason": "Checkpointing-in-eval and quantized offloading are unrelated bugs." + "reason": "Gradient checkpointing in eval mode and quantized offloading are unrelated behaviors with no shared code path." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation drift vs docs still using pipeline() are different issues." + "reason": "SigLIP2 implementation discrepancy and docs still using pipeline() are different bug classes." }, { "left": "issue:33290", "right": "issue:35141", "accept": false, - "reason": "Deepspeed Adafactor OOM is unrelated to embedding reinitialization." + "reason": "AdaFactor/DeepSpeed OOM and embedding reinitialization after resize are unrelated training issues." }, { "left": "issue:33290", "right": "issue:34689", "accept": false, - "reason": "Optimizer OOM and model-loading regression are different failure modes." + "reason": "Optimizer memory usage and model-loading regression are different problems in different parts of the stack." }, { "left": "issue:44077", "right": "issue:44479", "accept": false, - "reason": "PatchTSMixer post_init policy and Qwen video regression do not share a code path." + "reason": "Optional post_init on patchtsmixer and video-input regression for Qwen variants are not the same bug." }, { - "left": "issue:41628", - "right": "issue:45372", + "left": "issue:43720", + "right": "issue:44442", "accept": false, - "reason": "Different missing imports from different APIs/packages." + "reason": "BitNet packed-weight loading and FastSpeech2Conformer tokenizer import/load failure are unrelated." }, { - "left": "issue:43720", - "right": "issue:44442", + "left": "issue:41628", + "right": "issue:45372", "accept": false, - "reason": "BitNet accelerate loading and tokenizer loading are unrelated." + "reason": "Both involve imports, but they are different missing symbols from different optional dependencies and processor paths." }, { "left": "issue:43749", "right": "issue:43931", "accept": false, - "reason": "FSDP CPU-efficient loading and Qwen3-VL shape mismatch are separate loader bugs." + "reason": "FSDP CPU-efficient loading and Qwen3-VL weight-shape mismatch are distinct loading failures." }, { "left": "issue:43643", "right": "issue:44913", "accept": false, - "reason": "trust_remote_code field loss is not the same as rotary_pct reset on reload." + "reason": "Missing fields with trust_remote_code and GPTNeoX rotary_pct resetting on reload are separate config bugs." }, { "left": "issue:43873", "right": "issue:44871", "accept": false, - "reason": "Quantized offloading and Gemma eos_token_id mismatch are unrelated." + "reason": "Quantization/offloading behavior and Gemma-3 eos_token_id inconsistency are unrelated." }, { "left": "issue:39692", "right": "issue:44206", "accept": false, - "reason": "Docs example errors are not the same as a center-arg crash in feature extraction." + "reason": "SigLIP2 docs/example problems and LasrFeatureExtractor center-arg regression are different issues." }, { "left": "issue:44190", "right": "issue:44496", "accept": false, - "reason": "Local dataset loading and unrecognized model config are different problems." + "reason": "Local dataset loading in an example script and unrecognized model/config loading are unrelated." }, { - "left": "issue:33290", - "right": "issue:36331", + "left": "issue:43381", + "right": "issue:43856", "accept": false, - "reason": "OOM in training versus a Trainer signature mismatch." + "reason": "Checkpointing-in-eval and Qwen3 MoE memory usage are different training/runtime issues." }, { - "left": "issue:43381", - "right": "issue:43856", + "left": "issue:33290", + "right": "issue:36331", "accept": false, - "reason": "Eval-mode checkpointing and MoE memory inefficiency are separate issues." + "reason": "DeepSpeed OOM and CustomTrainer compute_loss signature breakage are not the same code-path problem." }, { "left": "issue:34689", "right": "issue:36010", "accept": false, - "reason": "Model-loading regression and GenerationMixin import breakage are different." + "reason": "Llama 3.2 vision loading regression and GenerationMixin import-path break are distinct import/load regressions." }, { "left": "issue:44855", "right": "issue:45137", "accept": false, - "reason": "Python 3.13 script parsing bug and ZeRO-3 deque error are unrelated." + "reason": "Python 3.13 IndentationError in DebertaV2Model and ZeRO3 deque pop error are unrelated failures." }, { "left": "issue:43540", "right": "issue:44514", "accept": false, - "reason": "Video input validation in Qwen3OmniMoe differs from batched chat-template padding crash." + "reason": "Both touch multimodal inputs, but Qwen3Omni video processing and Qwen2_5_VL batched chat-template padding failure are different bugs." }, { "left": "issue:43316", "right": "issue:44877", "accept": false, - "reason": "Gemma3TextConfig API mismatch is not the same as granite_speech strict-config loading." + "reason": "Gemma3TextConfig API mismatch and granite_speech strict-config loading are separate config issues." }, { "left": "issue:44162", "right": "issue:44792", "accept": false, - "reason": "ESM2 breakage and a janus test failure are unrelated." + "reason": "ESM2 breakage and janus image-generation test failure are unrelated model-specific issues." }, { "left": "issue:43388", "right": "issue:43688", "accept": false, - "reason": "Label truncation in gather_for_metrics is different from aux-loss normalization." + "reason": "Label dropping in gather_for_metrics and auxiliary-loss normalization are different training/evaluation code paths." }, { "left": "issue:42175", "right": "issue:43316", "accept": false, - "reason": "Missing TensorFlow backend and config API discrepancy are unrelated." + "reason": "Packaging backend omission and Gemma3TextConfig API discrepancy are unrelated." }, { "left": "issue:16998", "right": "issue:22355", "accept": false, - "reason": "model_max_length question and missing transformers.onnx import are unrelated." + "reason": "Model_max_length question for DeBERTa-V3 and missing transformers.onnx import are unrelated." }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "lm_head tying bug and FlashAttention failure are different Mistral-3 issues." + "reason": "Mistral3 lm_head tying and Ministral-3 FlashAttention failure are different model-specific bugs." }, { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Pipeline output shape change is unrelated to missing PyTorch." + "reason": "Pipeline top_k output shape change and missing PyTorch import are unrelated." }, { "left": "issue:43065", "right": "issue:44617", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as Sam3Video OOM." - }, - { - "left": "issue:41950", - "right": "issue:43475", - "accept": false, - "reason": "Image-processor lookup in video classification is unrelated to missing fpn embeddings." - }, - { - "left": "issue:43316", - "right": "issue:43716", - "accept": false, - "reason": "Gemma3TextConfig API discrepancy and Mistral3 dtype mismatch are different bugs." + "reason": "Sam3PixelDecoder dummy Conv2d and Sam3Video CUDA OOM are different issues; one is architecture scaffolding, the other memory pressure." } ] }, @@ -8070,11 +8109,13 @@ "nodes": [ "issue:32090", "issue:35141", - "issue:41720", + "issue:41950", "issue:42617", "issue:43065", "issue:43295", + "issue:43316", "issue:43452", + "issue:43475", "issue:43525", "issue:43531", "issue:43550", @@ -8082,6 +8123,7 @@ "issue:43606", "issue:43646", "issue:43701", + "issue:43716", "issue:43742", "issue:43761", "issue:43825", @@ -8104,20 +8146,19 @@ "issue:44871", "issue:44913", "issue:44933", - "issue:44938", "issue:44977", "issue:45127", "issue:45200", - "issue:45245", "issue:45290", "issue:45362", "issue:45381", "issue:45406", - "issue:45542", "issue:45563" ], "soft_pairs": [ + "issue:43316|issue:43716", "issue:43295|issue:43825", + "issue:41950|issue:43475", "issue:45127|issue:45200", "issue:43761|issue:44479", "issue:44077|issue:44534", @@ -8141,70 +8182,80 @@ "issue:43525|issue:43531", "issue:44792|issue:44977", "issue:44933|issue:45362", - "issue:43065|issue:44610", "issue:43827|issue:43994", - "issue:44534|issue:44913", - "issue:41720|issue:45245", - "issue:44938|issue:45542" + "issue:43065|issue:44610", + "issue:44534|issue:44913" ], - "prepared_review_unit_hash": "cc4c1bd2113bad04c3324d0d70cec78dcdb18f8bdd11b6f2f37d400b65f5e0e3", + "prepared_review_unit_hash": "689ee891aaf0fabad24ab24e55c21b4e56fae00df273e09a015d9a29af1a2c79", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35701, - "estimated_input_tokens": 8926, - "estimated_eval_tokens": 18108 + "serialized_chars": 35697, + "estimated_input_tokens": 8925, + "estimated_eval_tokens": 18106 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are a mixed set of unrelated bugs and regressions across loading, generation, multimodal chat/template handling, quantization/offload, and model-specific config issues. The soft edges mostly connect superficially similar reports, but they do not appear to be the same underlying bug or change.", - "confidence": 0.93, - "canonical_issue_reason": "issue:44521 is a clear, active, well-scoped bug report with multiple references and a concrete multimodal failure mode.", + "summary": "This cluster is highly heterogeneous: the items span unrelated bugs in model loading, chat templates, multimodal processors, generation, config serialization, offloading, and serving. I did not find any soft pair that clearly refers to the same underlying bug/change.", + "confidence": 0.94, + "canonical_issue_reason": "No strong canonical issue exists because the cluster is not a single duplicate family. If one issue must be chosen, issue 44521 is the broadest user-facing regression and has the most cross-references.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44521 is one of the strongest triage candidates here: specific repro, active discussion, and an actionable code-path bug.", + "best_issue_reason": "issue 44521 is the most broadly impactful and actively referenced item in the set, with multiple inbound references and a cross-cutting multimodal template regression.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "Different subsystems and symptoms: Gemma3 config API mismatch vs Mistral-3 image preprocessor dtype mismatch." + }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Different failures: processor/tokenizer API regression versus a pipeline error-message/docs issue." + "reason": "Both are v5 regressions, but one is processor/tokenizer access and image passing; the other is a pipeline error message about translation support." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Both involve video/image models, but the concrete failures differ: wrong processor lookup vs missing fpn_position_embeddings attribute." }, { "left": "issue:45127", "right": "issue:45200", "accept": false, - "reason": "Different models and mechanisms: LoRA merge/tied embeddings collapse versus Gemma 4 mm_token_type_ids defaults." + "reason": "Unrelated bugs: LoRA merge with extended vocab causes collapse, while Gemma 4 text-only fine-tuning needs default mm_token_type_ids." }, { "left": "issue:43761", "right": "issue:44479", "accept": false, - "reason": "Different model families and code paths: CLIP hidden_states regression versus Qwen VL video-input regression." + "reason": "Different model families and failure modes: CLIPVision hidden_states regression vs Qwen VL video-input regression." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "Optional post_init allowance for patchtsmixer is unrelated to non-persistent buffer initialization in v5." + "reason": "One is about optional post_init in patchtsmixer; the other is a v5 non-persistent buffer initialization problem." }, { "left": "issue:43550", "right": "issue:43701", "accept": false, - "reason": "Unrelated: torch.compile/SDPA failure in Bamba versus resume_from_checkpoint key mismatch." + "reason": "Torch.compile SDPA failure for Bamba is unrelated to resume_from_checkpoint key mismatches." }, { "left": "issue:43295", @@ -8216,139 +8267,127 @@ "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "CPU offload device mismatch for bark-small is unrelated to checkpoint key mismatch." + "reason": "CPU offload device mismatch is not the same bug as checkpoint resume key mismatch." }, { "left": "issue:44734", "right": "issue:45406", "accept": false, - "reason": "Both involve serving, but one is KV-cache indexing and the other is a missing processor attribute; different bugs." + "reason": "Both are serve-related, but one is KV-cache tensor indexing and the other is missing _tokenizer on Gemma4Processor." }, { "left": "issue:43646", "right": "issue:44913", "accept": false, - "reason": "Custom model init breakage is unrelated to GPTNeoX rotary_pct reload behavior." + "reason": "Generic custom model init breakage is not the same as GPTNeoX rotary_pct not persisting on reload." }, { "left": "issue:43452", "right": "issue:44496", "accept": false, - "reason": "GGUF loading breakage is not the same as an unrecognized config/model_type error." + "reason": "Both are model-loading failures, but one is gguf_file/from_pretrained behavior and the other is a missing model_type in config.json." }, { "left": "issue:43906", "right": "issue:45290", "accept": false, - "reason": "Insufficient evidence of same root cause; one is an isolated repro of another issue, the other is a specific chat-template crash." + "reason": "Different chat/template errors: isolated reproduction of an older issue vs assistant messages with tool calls and no content." }, { "left": "issue:43575", "right": "issue:43742", "accept": false, - "reason": "OOM during tensor-parallel loading is unrelated to a key error on a different model." + "reason": "OOM on large Qwen2 model with tensor parallelism is unrelated to a key error loading MobileLLM-125M." }, { "left": "issue:43761", "right": "issue:44242", "accept": false, - "reason": "CLIP hidden_states regression and load-balancing loss omission are different subsystems and symptoms." + "reason": "Hidden_states regression in CLIPVision is unrelated to router-logits/load-balancing loss behavior." }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "3d_parallel.py startup failure is unrelated to Qwen2 tensor-parallel OOM." + "reason": "3d_parallel.py execution failure is not the same concrete issue as Qwen2 loading OOM under tp." }, { "left": "issue:43931", "right": "issue:44060", "accept": false, - "reason": "Weight shape mismatch on load is different from a tied-weights warning bug." + "reason": "Weight-shape mismatch loading Qwen3-VL is distinct from the incorrect tied-weights warning in Qwen3-Next." }, { "left": "issue:44521", "right": "issue:45381", "accept": false, - "reason": "Both are multimodal regressions, but assistant masks in chat template and vision_position_ids in video input are different code paths." + "reason": "Both are multimodal regressions, but one is assistant mask generation and the other is video vision_position_ids." }, { "left": "issue:43575", "right": "issue:43873", "accept": false, - "reason": "TP OOM and quantization/offloading behavior are not the same underlying issue." + "reason": "Memory OOM under tensor parallelism is not the same as quantization/offloading not behaving as expected." }, { "left": "issue:32090", "right": "issue:35141", "accept": false, - "reason": "Trainer broadcast NoneType error is unrelated to embedding reinitialization after resize." + "reason": "Trainer broadcast TypeError and post_init embedding reinitialization are unrelated bugs." }, { "left": "issue:44464", "right": "issue:44871", "accept": false, - "reason": "Compiled chunked generation inconsistency is unrelated to eos_token_id config mismatch." + "reason": "Compiled-forward chunked generation inconsistency is unrelated to Gemma-3 eos_token_id configuration drift." }, { "left": "issue:44464", "right": "issue:45563", "accept": false, - "reason": "Generation inconsistency and a stale warning are different bugs." + "reason": "A generation correctness bug is not the same as a stale warning about num_return_sequences." }, { "left": "issue:44479", "right": "issue:44734", "accept": false, - "reason": "Video-input regression in Qwen VL is unrelated to serve KV-cache continuation indexing." + "reason": "Video input regression in Qwen VL is unrelated to KV-cache continuation indexing in serve." }, { "left": "issue:43525", "right": "issue:43531", "accept": false, - "reason": "Missing pad_token_id in Llama4Config is unrelated to Qwen3-MoE sliding_window behavior." + "reason": "Missing pad_token_id on Llama4Config is unrelated to Qwen3-MoE sliding_window behavior." }, { "left": "issue:44792", "right": "issue:44977", "accept": false, - "reason": "Janus image-generation test failure is unrelated to Qwen3.5 flash-attention generation issues." + "reason": "Janus image-generation test failure does not match Qwen3.5 flash-attention generation failure." }, { "left": "issue:44933", "right": "issue:45362", "accept": false, - "reason": "Missing import from image_utils is unrelated to a Qwen3.5 chat crash." - }, - { - "left": "issue:43065", - "right": "issue:44610", - "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder is unrelated to OmDet-Turbo processor/model input-size mismatch." + "reason": "Import error from image_utils is unrelated to Qwen3.5 chat crashes." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "Stale docs/error-message references to pipeline are not the same as SigLIP2 producing bad outputs." - }, - { - "left": "issue:44534", - "right": "issue:44913", - "accept": false, - "reason": "Non-persistent buffer junk in v5 is unrelated to GPTNeoX rotary_pct not persisting on reload." + "reason": "Docs still referencing pipeline() is not the same as SigLIP2 producing bad outputs with AutoModel/pipeline." }, { - "left": "issue:41720", - "right": "issue:45245", + "left": "issue:43065", + "right": "issue:44610", "accept": false, - "reason": "Qwen3 auto device-map cuda assert is unrelated to the category-count limit error." + "reason": "Dummy Conv2d in Sam3PixelDecoder is unrelated to OmDet-Turbo processor/model resolution mismatch." }, { - "left": "issue:44938", - "right": "issue:45542", + "left": "issue:44534", + "right": "issue:44913", "accept": false, - "reason": "Python 3.14 import/load failure is unrelated to a missing TensorFlow backend when only tensorboard is installed." + "reason": "Non-persistent buffer junk in v5 is a different failure mode from rotary_pct not surviving reload." } ] }, @@ -8377,8 +8416,8 @@ "issue:38617", "issue:39692", "issue:41628", + "issue:41720", "issue:41950", - "issue:42175", "issue:43065", "issue:43116", "issue:43295", @@ -8400,27 +8439,29 @@ "issue:44190", "issue:44220", "issue:44246", - "issue:44279", "issue:44355", "issue:44466", "issue:44496", - "issue:44683", "issue:44734", "issue:44743", "issue:44855", "issue:44871", "issue:44913", + "issue:44938", "issue:44964", "issue:45003", "issue:45127", "issue:45216", + "issue:45245", "issue:45276", "issue:45357", "issue:45372", "issue:45375", - "issue:45446" + "issue:45542" ], "soft_pairs": [ + "issue:41720|issue:45245", + "issue:44938|issue:45542", "issue:44190|issue:44964", "issue:43388|issue:43582", "issue:43650|issue:44220", @@ -8441,224 +8482,209 @@ "issue:43644|issue:44871", "issue:41950|issue:44016", "issue:43479|issue:45375", - "issue:44186|issue:44734", "issue:36246|issue:41628", - "issue:38175|issue:43065", + "issue:44186|issue:44734", "issue:44466|issue:44913", + "issue:38175|issue:43065", "issue:36246|issue:38617", - "issue:44246|issue:45003", - "issue:44683|issue:45446", - "issue:44016|issue:44220", - "issue:42175|issue:44279" + "issue:44246|issue:45003" ], - "prepared_review_unit_hash": "708a6ec2d626ff58f00e173d2efbec7f96e8083997460b229c118f366e0d968c", + "prepared_review_unit_hash": "896ab4dbeee29e631fce75c5df075bf4efd799d818849c0171233dad0694d671", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 36158, - "estimated_input_tokens": 9040, - "estimated_eval_tokens": 18336 + "soft_pair_count": 28, + "serialized_chars": 35881, + "estimated_input_tokens": 8971, + "estimated_eval_tokens": 18198 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is a loose collection of unrelated Transformers bugs and regressions. Despite some superficial overlap around imports, configs, multimodal models, and flex_attention, none of the soft pairs look like the same underlying issue, so I would not merge any of them.", - "confidence": 0.87, - "canonical_issue_reason": "issue:41628 is the cleanest representative of the set: a clear, user-facing core API import regression with a concise title and broad impact.", + "summary": "This set is mostly heterogeneous and does not form one real duplicate cluster. The only close duplicate-like thread is the Qwen3.5 save_pretrained regression (issue 45216, with follow-up 45357); all proposed soft edges are false positives.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 45216 is the best canonical among the set because it is the broadest, earliest report of the Qwen3.5 save_pretrained regression, and issue 45357 reads like a later follow-up on the same checkpoint/visual-key corruption.", "canonical_pr_reason": null, - "best_issue_reason": "issue:41628 is the strongest standalone issue here because it is specific, high-signal, and broadly relevant to import breakage in the library.", + "best_issue_reason": "45216 is the clearest representative issue: concrete regression, explicit version boundary, and strong overlap with the later Qwen3.5 save_pretrained report.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Different failures in unrelated areas: Qwen3 auto device mapping CUDA assert vs category-count limit error." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python 3.14 import failure is unrelated to the TensorFlow/tensorboard backend issue." + }, { "left": "issue:44190", "right": "issue:44964", "accept": false, - "reason": "Different failures: local dataset loading in an example script vs loading a specific multimodal model; no shared code path." + "reason": "Different models and symptoms: local dataset loading for an example script vs loading Phi-4 multimodal." }, { "left": "issue:43388", "right": "issue:43582", "accept": false, - "reason": "Metric label dropping vs Apple Silicon allocator TypeError are unrelated bugs." + "reason": "Training metric batching bug is unrelated to Apple Silicon allocator warmup TypeError." }, { "left": "issue:43650", "right": "issue:44220", "accept": false, - "reason": "A placeholder/low-content issue vs an audio feature extraction bug; no common underlying defect." + "reason": "No clear shared bug; one is a placeholder/spam-like issue, the other is an audio feature extraction problem." }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "LoRA merge/tied-weights collapse vs incorrect visual encoder keys on save_pretrained are different serialization/model-editing problems." + "reason": "Both are Qwen-related, but the bugs differ: LoRA/extended-vocab collapse vs save_pretrained visual-key serialization." }, { "left": "issue:43761", "right": "issue:44743", "accept": false, - "reason": "CLIPVision hidden_states regression vs recurrent state reset in Qwen3.5 modular code; different model paths." + "reason": "Different model families and code paths: CLIPVision hidden_states vs Qwen recurrent cache reset." }, { "left": "issue:43644", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer initialization bug vs GPTNeoX config reload defaulting bug are distinct issues." + "reason": "Generic v5 buffer serialization bug vs GPTNeoXConfig rotary_pct reload regression are unrelated." }, { "left": "issue:44077", "right": "issue:44079", "accept": false, - "reason": "Config validation around post_init vs ModelOutput key assignment behavior; not the same bug." + "reason": "Separate core bugs: patchtsmixer post_init validation vs ModelOutput key assignment." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Running compiled Python files vs DebertaV2 import parsing on Python 3.13 are unrelated." + "reason": "Compiled Python-file errors are not the same as the DebertaV2 Python 3.13 IndentationError." }, { "left": "issue:43976", "right": "issue:44855", "accept": false, - "reason": "General Python-version incompatibility vs a specific torch.jit/comment parsing error in one model." + "reason": "Version compatibility failure is unrelated to the decorator/comment parsing bug in DebertaV2." }, { "left": "issue:43540", "right": "issue:44734", "accept": false, - "reason": "Video input processing in Qwen3OmniMoe vs KV-cache continuation indexing in transformers serve are different code paths." + "reason": "Video input processing in Qwen3OmniMoe is unrelated to serve KV-cache tensor indexing." }, { "left": "issue:44496", "right": "issue:45372", "accept": false, - "reason": "Unrecognized model/config issue vs Gemma 4 processor import failure; different root causes." + "reason": "Unrecognized model/config error is unrelated to the mistral_common ReasoningEffort import breakage." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "Processor.tokenizer regression vs Gemma4 embedding resize behavior; both multimodal-adjacent but not the same defect." + "reason": "Processor/tokenizer regression is unrelated to resize_token_embeddings not updating Gemma4 embeddings." }, { "left": "issue:43550", "right": "issue:44190", "accept": false, - "reason": "Bamba torch.compile/SDPA failure is unrelated to the image-classification local dataset loading problem." + "reason": "torch.compile/SDPA failure is unrelated to the image-classification no-trainer dataset loading problem." }, { "left": "issue:43335", "right": "issue:44855", "accept": false, - "reason": "SwitchTransformers sparse-layer config bug vs DebertaV2 import syntax/parsing bug; no overlap." + "reason": "SwitchTransformers sparse-layer config bug is unrelated to the Python 3.13 parsing/IndentationError." }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "LM head weight tying in Mistral3 vs incorrect Qwen3.5 save_pretrained keys are separate weight-serialization issues." + "reason": "Both involve tied weights/checkpointing, but they are different model families and different concrete bugs." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Multi-label classification empty results vs local dataset loading in another example script are different example-script bugs." + "reason": "Empty multi-label classification outputs are unrelated to local dataset loading in an example script." }, { "left": "issue:39692", "right": "issue:44190", "accept": false, - "reason": "SigLIP2 documentation example errors vs image-classification no-trainer dataset loading are unrelated." + "reason": "SigLIP2 documentation/model mismatch is unrelated to the no-trainer image-classification dataset issue." }, { "left": "issue:43644", "right": "issue:44871", "accept": false, - "reason": "Buffer initialization bug vs eos_token_id mismatch on Gemma-3 are separate config/model-state issues." + "reason": "Non-persistent buffer junk in v5 is unrelated to Gemma-3 eos_token_id config mismatch." }, { "left": "issue:41950", "right": "issue:44016", "accept": false, - "reason": "Video-classification pipeline processor lookup vs notebook syntax error are unrelated." + "reason": "Video-classification processor lookup bug is unrelated to a notebook syntax error." }, { "left": "issue:43479", "right": "issue:45375", "accept": false, - "reason": "Different multimodal config problems for different models/fields; not the same underlying bug." - }, - { - "left": "issue:44186", - "right": "issue:44734", - "accept": false, - "reason": "Tokenizer crash on NER/padding vs serve KV-cache tensor indexing are unrelated." + "reason": "Different config bugs in different multimodal models; not the same underlying issue." }, { "left": "issue:36246", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but one is a missing submodule export and the other is a missing top-level symbol; different API breakages." + "reason": "Both are import errors, but for different missing symbols and different API surfaces." }, { - "left": "issue:38175", - "right": "issue:43065", + "left": "issue:44186", + "right": "issue:44734", "accept": false, - "reason": "SigLIP2 zero probabilities vs Sam3PixelDecoder dummy Conv2d are unrelated model-behavior issues." + "reason": "LayoutLMv2 tokenizer NER/padding crash is unrelated to serve KV-cache continuation indexing." }, { "left": "issue:44466", "right": "issue:44913", "accept": false, - "reason": "Device-dependent lm_head serialization vs GPTNeoX rotary_pct reload defaulting are different config/serialization bugs." - }, - { - "left": "issue:36246", - "right": "issue:38617", - "accept": false, - "reason": "Different missing imports in different modules; same symptom, not same bug." - }, - { - "left": "issue:44246", - "right": "issue:45003", - "accept": false, - "reason": "Slow import vs unsafe sys.modules access may be related in theme, but the concrete problems are not the same." - }, - { - "left": "issue:44683", - "right": "issue:45446", - "accept": false, - "reason": "Compiled flex_attention failure on torch>=2.9 vs an incorrect version check for AuxRequest import are different flex_attention bugs." + "reason": "Inconsistent lm_head.weight serialization is unrelated to GPTNeoX rotary_pct reload behavior." }, { - "left": "issue:44016", - "right": "issue:44220", + "left": "issue:38175", + "right": "issue:43065", "accept": false, - "reason": "Notebook syntax error vs audio feature extraction issue are unrelated." + "reason": "SigLIP2 zero probabilities and Sam3PixelDecoder dummy Conv2d are different model-specific bugs." }, { - "left": "issue:42175", - "right": "issue:44279", + "left": "issue:36246", + "right": "issue:38617", "accept": false, - "reason": "TensorFlow missing from a [torch] install is not the same as a general dependency issue." + "reason": "Different import failures: missing Qwen2_5_VLImageProcessor vs missing layer_type_validation." }, { "left": "issue:44246", "right": "issue:45003", "accept": false, - "reason": "Import-time slowness and unsafe sys.modules access may touch import machinery, but they are not the same concrete defect." + "reason": "Import slowness is a performance complaint, not the sys.modules safety bug." } ] }, @@ -8683,6 +8709,7 @@ "cluster_id": "cluster-15354-366", "nodes": [ "issue:41950", + "issue:42175", "issue:42907", "issue:43295", "issue:43299", @@ -8702,23 +8729,21 @@ "issue:44062", "issue:44075", "issue:44112", + "issue:44220", "issue:44242", "issue:44279", "issue:44336", "issue:44351", "issue:44368", - "issue:44462", "issue:44496", "issue:44521", "issue:44683", "issue:44792", "issue:44871", - "issue:44877", "issue:44933", "issue:44964", "issue:44987", "issue:45042", - "issue:45081", "issue:45200", "issue:45216", "issue:45245", @@ -8727,11 +8752,15 @@ "issue:45335", "issue:45357", "issue:45405", + "issue:45446", "issue:45542", "issue:45561", "issue:45563" ], "soft_pairs": [ + "issue:44683|issue:45446", + "issue:44016|issue:44220", + "issue:42175|issue:44279", "issue:44112|issue:44792", "issue:43295|issue:45335", "issue:43643|issue:44496", @@ -8741,8 +8770,8 @@ "issue:44016|issue:44062", "issue:44112|issue:44964", "issue:44871|issue:45245", - "issue:43854|issue:44964", "issue:43673|issue:44336", + "issue:43854|issue:44964", "issue:43976|issue:44987", "issue:44496|issue:45042", "issue:43526|issue:43901", @@ -8758,212 +8787,216 @@ "issue:43976|issue:45405", "issue:44351|issue:45278", "issue:43479|issue:43525", - "issue:41950|issue:43650", - "issue:44496|issue:44877", - "issue:44462|issue:45081" + "issue:41950|issue:43650" ], - "prepared_review_unit_hash": "d2831091fceab457522878ab18643d7f28cb77089cda4264f6e33018a9f271ff", + "prepared_review_unit_hash": "d4fd8bd9aee055224ac3b07d063722a2abfb5f6e01c68afe195e57fd75d6db0a", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36379, - "estimated_input_tokens": 9095, - "estimated_eval_tokens": 18446 + "soft_pair_count": 30, + "serialized_chars": 36583, + "estimated_input_tokens": 9146, + "estimated_eval_tokens": 18548 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is dominated by loosely related Transformers regressions around model loading, config handling, tokenizers, and multimodal processing. The soft pairs mostly share broad subsystem keywords but not the same concrete bug, so none should be merged as duplicates.", - "confidence": 0.31, - "canonical_issue_reason": "Issue 44964 is an open, concrete latest-transformers model-loading regression and is representative of the cluster\u2019s main theme better than the narrower or more one-off reports.", + "summary": "This cluster is mostly heterogeneous Transformers issues with only superficial similarity between a few pairs. I would not merge any of the soft-edge candidates; they share broad subsystems or model families, but not the same concrete bug/change.", + "confidence": 0.46, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella-style report in the set, covering the import-regression theme seen in several nearby issues, but the cluster is still too mixed for a perfect canonical issue.", "canonical_pr_reason": null, - "best_issue_reason": "44964 is the best anchor issue because it is current, user-facing, and specific enough to represent the broader family of load/config regressions in this set.", + "best_issue_reason": "45278 is the best representative issue because it captures the widest shared symptom class (post-upgrade import breakage) among the otherwise scattered reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44112", - "right": "issue:44792", + "left": "issue:44683", + "right": "issue:45446", "accept": false, - "reason": "Both are test failures, but they concern different models and different failure modes." + "reason": "Both touch flex_attention, but one is a torch>=2.9 compilation failure and the other is a version-check bug for AuxRequest import; different root causes." }, { - "left": "issue:43295", + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "A notebook syntax error and a feature-extraction function issue are unrelated bugs." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "One is a missing TensorFlow backend in the torch extra, the other is a generic dependency complaint; not the same concrete defect." + }, + { + "left": "issue:44112", + "right": "issue:44792", + "accept": false, + "reason": "Different models and different failure modes: CI stale test in GraniteSpeech vs janus image-generation test failure." + }, + { + "left": "issue:43295", "right": "issue:45335", "accept": false, - "reason": "One is a processor/tokenizer regression; the other is an embedding resize bug in t5gemma. Different code paths." + "reason": "Both are regressions, but one is about processor/tokenizer image handling and the other about resize_token_embeddings not updating decoder.embed_tokens." }, { "left": "issue:43643", "right": "issue:44496", "accept": false, - "reason": "Both involve config loading, but one is missing fields with trust_remote_code and the other is an unrecognized-model/config.json issue." + "reason": "Both involve config loading, but one is missing fields with trust_remote_code and the other is an unrecognized model_type error." }, { "left": "issue:45561", "right": "issue:45563", "accept": false, - "reason": "A pytest-xdist file race and a stale generate warning are unrelated bugs." + "reason": "A pytest-xdist file race and a stale generation warning are unrelated issues." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Both mention Qwen3.5/Qwen3VL, but one is model loading and the other is incorrect save_pretrained output." + "reason": "Same model family, but loading a Qwen3.5 MoE model and save_pretrained saving wrong keys are distinct code paths." }, { "left": "issue:42907", "right": "issue:43441", "accept": false, - "reason": "Both are Ministral-related, but one is saving dequantized weights and the other is FlashAttention loading failure." + "reason": "Different Ministral failures: dequantized save path vs FlashAttention inference failure." }, { "left": "issue:44016", "right": "issue:44062", "accept": false, - "reason": "Notebook syntax error versus AddedToken argument collision; no shared underlying bug." + "reason": "Notebook syntax error vs AddedToken keyword-argument error; no shared underlying bug." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "GraniteSpeech CI device override failure and Phi-4 multimodal loading are distinct regressions." + "reason": "CI test flake in GraniteSpeech is unrelated to Phi-4 multimodal loading." }, { "left": "issue:44871", "right": "issue:45245", "accept": false, - "reason": "EOS token config mismatch and category-count limits are unrelated." + "reason": "Gemma eos_token_id inconsistency and a category-count runtime error are unrelated." }, { - "left": "issue:43854", - "right": "issue:44964", + "left": "issue:43673", + "right": "issue:44336", "accept": false, - "reason": "Both are model loading issues, but for different models and different root causes." + "reason": "Chunked prefill cache handling and ANSI output in loading_report do not describe the same bug." }, { - "left": "issue:43673", - "right": "issue:44336", + "left": "issue:43854", + "right": "issue:44964", "accept": false, - "reason": "Generation cache handling and ANSI output formatting are unrelated." + "reason": "Both are model-loading failures, but for different models and likely different causes." }, { "left": "issue:43976", "right": "issue:44987", "accept": false, - "reason": "Both mention version-related breakage, but one is Python support and the other is a specific model load failure." + "reason": "A Python version support issue and a model-loading regression are not duplicates." }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Unrecognized-model config error versus PIL image processor torchvision dependency bug; different subsystems." + "reason": "Unrecognized model config vs PIL image-processor torchvision dependency are unrelated." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "BeitImageProcessorFast label reduction and pipeline docs drift are not the same bug." + "reason": "A processor bug and a documentation mismatch are not the same issue." }, { "left": "issue:43824", "right": "issue:44933", "accept": false, - "reason": "Qwen2.5-VL import failure and image_utils import failure are separate missing-symbol issues." + "reason": "Both are import-related, but they reference different symbols and different failure contexts." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "A Python version incompatibility and a generic dependency issue are not duplicates." + "reason": "Python compatibility and dependency installation issues are not the same bug." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "Both involve Qwen3.5 save/load regressions, but they affect different artifacts: loading versus saved visual encoder keys." + "reason": "Both concern Qwen3.5 regressions, but loading failure and incorrect saved visual encoder keys are distinct problems." }, { "left": "issue:43577", "right": "issue:45216", "accept": false, - "reason": "BLIP2 dtype propagation and Qwen3.5 save_pretrained corruption are unrelated." + "reason": "Blip2 dtype loading and Qwen3.5 checkpoint serialization are unrelated." }, { "left": "issue:44075", "right": "issue:44368", "accept": false, - "reason": "Optimizer argument handling and a tie_word_embeddings warning are different concerns." + "reason": "Optimizer arguments being ignored and a tie_word_embeddings warning are unrelated." }, { "left": "issue:44521", "right": "issue:45325", "accept": false, - "reason": "Multimodal assistant mask generation and Qwen2.5-VL rope indexing are separate multimodal bugs." + "reason": "Both are multimodal, but assistant mask generation and rope-index scaling are separate code paths." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "A docs/behavior mismatch for return_all_scores is unrelated to config recognition failures." + "reason": "Documentation about return_all_scores does not duplicate an unrecognized-model loading error." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "MoE load-balancing loss and Gemma 4 token_type_id defaults are different training-path problems." + "reason": "MoE load-balancing loss and Gemma 4 mm_token_type_ids are different training/config bugs." }, { "left": "issue:44683", "right": "issue:45542", "accept": false, - "reason": "Compiled flex_attention on torch 2.9 and missing TensorFlow backend handling are unrelated." + "reason": "FlexAttention on torch>=2.9 and a missing TensorFlow backend are unrelated." }, { "left": "issue:43976", "right": "issue:45405", "accept": false, - "reason": "Python support and an unreleased PEFT minimum version bump are different release issues." + "reason": "Python support and a PEFT version bump are not the same issue." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "A missing HybridCache import and broad import errors after upgrade are too different to treat as the same bug." + "reason": "Both mention imports, but 45278 is a broad upgrade report and 44351 is a specific HybridCache import failure; not enough evidence of one root cause." }, { "left": "issue:43479", "right": "issue:43525", "accept": false, - "reason": "Phi4MultimodalConfig default initialization and Llama4Config missing pad_token_id are distinct config bugs." + "reason": "Different config classes and different missing-default behavior; not duplicates." }, { "left": "issue:41950", "right": "issue:43650", "accept": false, - "reason": "A video-classification processor lookup bug and a spammy placeholder issue have no meaningful overlap." - }, - { - "left": "issue:44496", - "right": "issue:44877", - "accept": false, - "reason": "Both are config-loading problems, but one is missing model_type and the other is strict granite_speech config validation." - }, - { - "left": "issue:44462", - "right": "issue:45081", - "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json and Mistral regex patch crashes are different tokenizer-loading failures." + "reason": "A pipeline processor lookup bug and a placeholder issue title are unrelated." } ] }, @@ -8990,7 +9023,6 @@ "issue:38175", "issue:39692", "issue:41950", - "issue:42175", "issue:42994", "issue:43065", "issue:43122", @@ -9008,8 +9040,6 @@ "issue:43784", "issue:43854", "issue:43901", - "issue:43976", - "issue:43994", "issue:44016", "issue:44162", "issue:44261", @@ -9019,6 +9049,7 @@ "issue:44297", "issue:44462", "issue:44466", + "issue:44496", "issue:44521", "issue:44617", "issue:44623", @@ -9026,10 +9057,12 @@ "issue:44683", "issue:44749", "issue:44779", + "issue:44877", "issue:44898", "issue:44964", "issue:44987", "issue:45020", + "issue:45081", "issue:45276", "issue:45290", "issue:45397", @@ -9037,21 +9070,23 @@ "issue:45542" ], "soft_pairs": [ + "issue:44496|issue:44877", + "issue:44462|issue:45081", "issue:41950|issue:45020", "issue:43550|issue:44964", "issue:44297|issue:44462", "issue:44279|issue:45405", "issue:43299|issue:43854", "issue:43295|issue:44779", - "issue:38175|issue:44617", "issue:44462|issue:44779", + "issue:38175|issue:44617", "issue:39692|issue:43901", "issue:43650|issue:45290", "issue:44016|issue:44291", "issue:44466|issue:45276", - "issue:43262|issue:44521", "issue:43450|issue:44279", "issue:43723|issue:44162", + "issue:43262|issue:44521", "issue:43441|issue:44683", "issue:43065|issue:44898", "issue:42994|issue:43317", @@ -9062,213 +9097,198 @@ "issue:43784|issue:45542", "issue:44671|issue:44779", "issue:43673|issue:43761", - "issue:44987|issue:45542", - "issue:43901|issue:43994", - "issue:44297|issue:45397", - "issue:42175|issue:43976" + "issue:44987|issue:45542" ], - "prepared_review_unit_hash": "827bf85dbfc7a1bd073dff97f22b153bae68e446aafd755717283fb19c28f808", + "prepared_review_unit_hash": "bc74b5b1033f1addd791a28eaae5eb05227f927665fc6b11c7a1e975d7afda67", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36207, - "estimated_input_tokens": 9052, - "estimated_eval_tokens": 18360 + "soft_pair_count": 28, + "serialized_chars": 36047, + "estimated_input_tokens": 9012, + "estimated_eval_tokens": 18280 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes tokenizer, processor, model-loading, quantization, docs, dependency, and attention-backend regressions across different models. None of the proposed soft pairs look like the same underlying bug/change, so all should stay separate.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue fits this cluster: the items are about different code paths and different models/features, so treating any one as the duplicate target would be misleading.", + "summary": "The cluster is heterogeneous: mostly unrelated Transformers v5 regressions across tokenizers, chat templates, model loading, quantization, and docs. None of the soft edges look like true duplicates of the same underlying bug/change, so I reject all of them. The best representative issue is the broad remote_code loading regression.", + "confidence": 0.67, + "canonical_issue_reason": "Issue 45020 is the broadest, highest-signal representative of the set: it captures a wide class of recent-version model loading breakages around remote_code and is more central than the narrow, model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "No safe global best issue: the cluster does not converge on one underlying defect or fix, so there is no representative issue to anchor deduping.", + "best_issue_reason": "45020 is the best standalone anchor for this cluster because it is the most general regression report with the widest potential overlap to adjacent loading failures.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44496", + "right": "issue:44877", + "accept": false, + "reason": "Both are loading/config errors, but one is an unrecognized-model/config.json problem and the other is strict-config handling for granite_speech; different failure modes and fixes." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json is unrelated to the Mistral regex patch crash; different code paths and symptoms." + }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "One is a video-classification processor lookup bug; the other is a broad remote_code loading regression. Different subsystems and failure modes." + "reason": "Video-classification processor lookup is not the same as broad remote_code model-loading regressions." }, { "left": "issue:43550", "right": "issue:44964", "accept": false, - "reason": "torch.compile/SDPA on Bamba is unrelated to loading Phi-4 multimodal models." + "reason": "torch.compile/SDPA failure in Bamba is distinct from Phi-4 multimodal loading issues." }, { "left": "issue:44297", "right": "issue:44462", "accept": false, - "reason": "Both involve tokenizers, but one is save_pretrained writing the wrong tokenizer_class and the other is AutoTokenizer ignoring tokenizer.json. Different paths." + "reason": "Tokenizer save metadata mismatch is a serialization bug; AutoTokenizer ignoring tokenizer.json is a load-path selection bug." }, { "left": "issue:44279", "right": "issue:45405", "accept": false, - "reason": "A generic dependency issue is not the same as bumping MIN_PEFT_VERSION to an unreleased version." + "reason": "Generic dependency issue is unrelated to a PEFT minimum-version bump." }, { "left": "issue:43299", "right": "issue:43854", "accept": false, - "reason": "These are model-specific loading failures for different architectures and likely different causes." + "reason": "Two separate model-loading regressions for different models with no shared concrete code path evident." }, { "left": "issue:43295", "right": "issue:44779", "accept": false, - "reason": "Processor.tokenizer/image-passing regression is unrelated to Deepseek tokenization output changes." + "reason": "Processor/tokenizer API regression is unrelated to Deepseek tokenizer output changes." }, { "left": "issue:38175", "right": "issue:44617", "accept": false, - "reason": "SigLIP2 zero probabilities and Sam3Video CUDA OOM are unrelated model/runtime issues." - }, - { - "left": "issue:44462", - "right": "issue:44779", - "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json is a loading/selection bug, while Deepseek incorrect results are a tokenization behavior regression." + "reason": "SigLIP2 probability issue and Sam3Video OOM are unrelated model/runtime problems." }, { "left": "issue:39692", "right": "issue:43901", "accept": false, - "reason": "Both are docs-related, but one is SigLIP2 example errors and the other is a pipeline doc mismatch; not the same issue." + "reason": "Docs/example errors for SigLIP2 are not the same as TextClassificationPipeline doc mismatch." }, { "left": "issue:43650", "right": "issue:45290", "accept": false, - "reason": "Random placeholder text is unrelated to apply_chat_template crashing on tool-call assistant messages." + "reason": "Placeholder issue title is not related to chat-template tool-call crashes." }, { "left": "issue:44016", "right": "issue:44291", "accept": false, - "reason": "Notebook syntax error and init_empty_weights argument mismatch are unrelated." + "reason": "Notebook syntax error and init_empty_weights argument regression are unrelated." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Inconsistent lm_head.weight serialization is not the same as resize_token_embeddings not updating Gemma4 embedding/output fields." - }, - { - "left": "issue:43262", - "right": "issue:44521", - "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and multimodal assistant mask generation are different bugs in different template code." + "reason": "Both mention weights/embeddings, but one is serialization inconsistency and the other is resize_token_embeddings propagation; not the same bug." }, { "left": "issue:43450", "right": "issue:44279", "accept": false, - "reason": "Batched video processor shape handling is unrelated to a dependency/version problem." + "reason": "Batched video processor shape bug is unrelated to a dependency/version issue." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "AutoTokenizer.from_pretrained loading regression and ESM2 breakage are separate model-loading problems." + "reason": "Tokenizer loading regression and ESM2 model breakage are different issues." + }, + { + "left": "issue:43262", + "right": "issue:44521", + "accept": false, + "reason": "Both involve apply_chat_template, but one is default sampling rate and the other is assistant mask generation for multimodal inputs." }, { "left": "issue:43441", "right": "issue:44683", "accept": false, - "reason": "FlashAttention failure for Ministral-3 and compiled flex_attention failure on torch>=2.9 are distinct attention-backend issues." + "reason": "FlashAttention failure for Ministral-3 and compiled flex_attention failure on torch>=2.9 are different attention-stack bugs." }, { "left": "issue:43065", "right": "issue:44898", "accept": false, - "reason": "Sam3PixelDecoder's dummy Conv2d and Perceiver non-default-resolution failures are unrelated vision-model bugs." + "reason": "Sam3PixelDecoder dummy conv and Perceiver resize/interpolate failure are unrelated architecture-specific issues." }, { "left": "issue:42994", "right": "issue:43317", "accept": false, - "reason": "Quantized model saving failure and dequantized model offload loading failure are different quantization lifecycle problems." + "reason": "Quantized model saving failure and dequantized offload loading failure are related only at a high level, not the same underlying bug." }, { "left": "issue:44261", "right": "issue:44263", "accept": false, - "reason": "Missing rms_norm_eps in MLA q_a_layernorm and torch.split return values in GlmMoeDsaIndexer are not the same defect." + "reason": "MLA layernorm eps precision issue and GlmMoeDsaIndexer split behavior are unrelated." }, { "left": "issue:44623", "right": "issue:45397", "accept": false, - "reason": "Missing files on processor.save_pretrained is unrelated to gemma-4 zero3 from_pretrained." + "reason": "Missing processor.save_pretrained files and gemma-4 zero3 from_pretrained are separate save/load problems." }, { "left": "issue:43122", "right": "issue:43650", "accept": false, - "reason": "Tokenizer-version behavior change is unrelated to the placeholder 'ADD THE DATA' issue." + "reason": "Tokenizer behavior change across versions is unrelated to the placeholder 'ADD THE DATA' issue." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Processor save_pretrained missing files has no overlap with the Chinese report about data filtering slowdown." + "reason": "Processor save files missing and Chinese performance regression are unrelated." }, { "left": "issue:43784", "right": "issue:45542", "accept": false, - "reason": "Both are dependency/backend errors, but one is sentence-transformers importing nn and the other is missing TensorFlow backend installation; different causes." + "reason": "Missing nn import in sentence-transformers and undefined tf backend when only tensorboard is installed are different optional-dependency failures." }, { "left": "issue:44671", "right": "issue:44779", "accept": false, - "reason": "CamemBERT masked LM prediction regression is unrelated to Deepseek tokenizer behavior." + "reason": "Masked LM prediction regression and Deepseek tokenizer regression affect different components and code paths." }, { "left": "issue:43673", "right": "issue:43761", "accept": false, - "reason": "Generation cache missing in chunked_prefill and CLIPVisionModel hidden_states handling are different forward-path regressions." + "reason": "Generation cache missing in chunked_prefill and CLIPVision hidden_states regression are unrelated model behaviors." }, { "left": "issue:44987", "right": "issue:45542", "accept": false, - "reason": "Loading a specific model on transformers>=5.1.0 is unrelated to tensorboard/TensorFlow backend installation." - }, - { - "left": "issue:43901", - "right": "issue:43994", - "accept": false, - "reason": "A documentation mismatch about return_all_scores is not the same as SigLIP2 producing nonsensical outputs." - }, - { - "left": "issue:44297", - "right": "issue:45397", - "accept": false, - "reason": "Tokenizer save metadata mismatch is unrelated to gemma-4 zero3 loading." - }, - { - "left": "issue:42175", - "right": "issue:43976", - "accept": false, - "reason": "Missing TensorFlow in the torch extra and Python version support in 5.1.0 are separate packaging issues." + "reason": "Physical-intelligence/fast loading failure is unrelated to the TensorFlow backend dependency issue." } ] }, @@ -9295,6 +9315,7 @@ "issue:36246", "issue:41628", "issue:41950", + "issue:42175", "issue:42491", "issue:42757", "issue:43295", @@ -9315,9 +9336,9 @@ "issue:44162", "issue:44188", "issue:44220", - "issue:44242", "issue:44246", "issue:44295", + "issue:44297", "issue:44336", "issue:44351", "issue:44373", @@ -9328,7 +9349,6 @@ "issue:44589", "issue:44617", "issue:44661", - "issue:44857", "issue:44871", "issue:45003", "issue:45200", @@ -9341,13 +9361,16 @@ "issue:45561" ], "soft_pairs": [ + "issue:43901|issue:43994", + "issue:44297|issue:45397", + "issue:42175|issue:43976", "issue:43475|issue:44016", "issue:43824|issue:43976", "issue:43352|issue:45250", "issue:43295|issue:44617", - "issue:43450|issue:44617", "issue:43525|issue:43881", "issue:44589|issue:45362", + "issue:43450|issue:44617", "issue:43901|issue:44246", "issue:44188|issue:44336", "issue:41628|issue:43976", @@ -9367,205 +9390,216 @@ "issue:43673|issue:44561", "issue:44336|issue:44485", "issue:44162|issue:45200", - "issue:44246|issue:44871", - "issue:44242|issue:44857" + "issue:44246|issue:44871" ], - "prepared_review_unit_hash": "30c9919fb8776d137f31ee767ac4507a8647e05a7405c3aee19c89006ba130a0", + "prepared_review_unit_hash": "2967174b8a6a71d9e66034282ad161cf8d29ca4a2799f5f9411e7614011313d7", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 28, - "serialized_chars": 35218, - "estimated_input_tokens": 8805, - "estimated_eval_tokens": 17866 + "soft_pair_count": 30, + "serialized_chars": 35861, + "estimated_input_tokens": 8966, + "estimated_eval_tokens": 18188 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is broadly over-linked by superficial similarity (mostly import/version/model-loading regressions). I don\u2019t see any soft pair that clearly shares the same underlying bug or change, so I would not merge any of the candidates as duplicates.", - "confidence": 0.63, - "canonical_issue_reason": "issue:43824 is the strongest representative: it\u2019s a concrete, user-facing import failure, has the highest discussion/inbound activity among the import-error subset, and is specific enough to anchor the cluster better than the more generic or unrelated reports.", + "summary": "This cluster is heterogeneous: most pairs share a broad subsystem or release theme, but they do not look like the same underlying bug or change. I would not collapse any of the soft edges, and there is no clear canonical issue/PR for the whole set.", + "confidence": 0.95, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43824 is the best overall issue candidate because it is specific, well-scoped, and clearly a real regression/import breakage with strong activity signals; the other items are either unrelated, narrower, or more ambiguous.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Both concern text-classification behavior, but one is a docs/behavior mismatch and the other is a model/pipeline correctness bug." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save metadata mismatch is unrelated to the Gemma-4 ZeRO-3 from_pretrained failure." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Different failures: missing TensorFlow backend in a torch install vs Python version compatibility breakage." + }, { "left": "issue:43475", "right": "issue:44016", "accept": false, - "reason": "Different problems: a SAM 3 video model attribute error vs a notebook syntax error." + "reason": "SAM3 video attribute error and a notebook syntax error are unrelated." }, { "left": "issue:43824", "right": "issue:43976", "accept": false, - "reason": "Both are import-related, but one is a missing Qwen2.5 VL symbol and the other is a Python-version compatibility report; not the same bug." + "reason": "Import error for a missing model class is unrelated to Python 3.9/3.10 packaging support." }, { "left": "issue:43352", "right": "issue:45250", "accept": false, - "reason": "One is a model-specific Flash Attention support error; the other is a generic FA2 topic, not the same concrete issue." + "reason": "One is a model-specific Flash Attention 2 incompatibility, the other is a generic Flash Attention topic." }, { "left": "issue:43295", "right": "issue:44617", "accept": false, - "reason": "Custom processor/tokenizer regression vs SAM 3 video OOM; unrelated failure modes." - }, - { - "left": "issue:43450", - "right": "issue:44617", - "accept": false, - "reason": "Incorrect batched video processor shapes vs CUDA memory exhaustion; different root causes." + "reason": "Processor/tokenizer regression is unrelated to the Sam3Video CUDA OOM report." }, { "left": "issue:43525", "right": "issue:43881", "accept": false, - "reason": "Different model families and different load failures; no shared code-path bug is evident." + "reason": "Missing pad_token_id on Llama4Config and glm-4v-9b loading failure are different model-loading bugs." }, { "left": "issue:44589", "right": "issue:45362", "accept": false, - "reason": "Float8 storage deserialization error vs Qwen3.5 chat crash; not the same underlying issue." + "reason": "Float8 storage lookup failure is unrelated to Qwen3.5-35B chat crashing." + }, + { + "left": "issue:43450", + "right": "issue:44617", + "accept": false, + "reason": "Incorrect batched video processor shape and CUDA OOM are distinct problems." }, { "left": "issue:43901", "right": "issue:44246", "accept": false, - "reason": "Docs mismatch about return_all_scores vs slow import performance; unrelated." + "reason": "Docs mismatch in text-classification pipeline is not the same as intermittent slow import." }, { "left": "issue:44188", "right": "issue:44336", "accept": false, - "reason": "Torch.compile attention-kernel divergence vs ANSI codes in loading_report; unrelated." + "reason": "Torch.compile kernel divergence and ANSI logging output are unrelated." }, { "left": "issue:41628", "right": "issue:43976", "accept": false, - "reason": "Missing AutoImageProcessor import vs Python compatibility breakage; too broad and not the same bug." + "reason": "Missing AutoImageProcessor import and Python-version incompatibility are different packaging issues." }, { "left": "issue:44295", "right": "issue:45397", "accept": false, - "reason": "position_ids buffer access error vs Gemma-4 Zero3 from_pretrained issue; different code paths." + "reason": "position_ids buffer read error is unrelated to Gemma-4 ZeRO-3 loading." }, { "left": "issue:42491", "right": "issue:43299", "accept": false, - "reason": "Both mention Qwen3 MoE/version changes, but one is LoRA reuse across versions and the other is model loading for Qwen3VL MoE; not clearly the same defect." + "reason": "Both mention Qwen3 MoE/VL, but the reported failures are different model/version compatibility issues." }, { "left": "issue:41950", "right": "issue:44220", "accept": false, - "reason": "Video-classification pipeline looking for image processors vs a feature-extraction bug; different subsystems and symptoms." + "reason": "Video classification using image processors is unrelated to _torch_extract_fbank_features()." }, { "left": "issue:44336", "right": "issue:44373", "accept": false, - "reason": "Terminal ANSI output issue vs wrong docstring for position_ids; unrelated." + "reason": "ANSI code leakage in loading_report and a wrong docstring are not the same bug." }, { "left": "issue:44246", "right": "issue:44496", "accept": false, - "reason": "Slow import performance vs unrecognized model/config; different failures." + "reason": "Slow import and unrecognized model config are unrelated failures." }, { "left": "issue:44462", "right": "issue:45356", "accept": false, - "reason": "Tokenizer.json being ignored vs Kimi-K2.5 tokenizer regression; both tokenizer-related but not the same concrete bug." + "reason": "AutoTokenizer ignoring tokenizer.json is not the same as the Kimi-K2.5 codec regression." }, { "left": "issue:44661", "right": "issue:45542", "accept": false, - "reason": "Model-registration edge case in tokenizer mapping vs tensorboard/TensorFlow backend issue; unrelated." + "reason": "TOKENIZER_MAPPING_NAMES failure and undefined tf backend from tensorboard-only install are different dependency issues." }, { "left": "issue:45003", "right": "issue:45561", "accept": false, - "reason": "Unsafe sys.modules access in modeling_utils vs pytest-xdist race on captured_info.txt; unrelated." + "reason": "Unsafe sys.modules access and pytest-xdist file race are unrelated implementation bugs." }, { "left": "issue:43994", "right": "issue:45276", "accept": false, - "reason": "Siglip2 inference mismatch vs Gemma-4 embedding resize inconsistency; different models and code paths." + "reason": "Siglip2 nonsensical outputs and Gemma-4 embedding resize propagation are different model behaviors." }, { "left": "issue:43526", "right": "issue:45003", "accept": false, - "reason": "BEiT image processor label reduction bug vs sys.modules access bug; unrelated." + "reason": "BeitImageProcessorFast reduce_labels bug and sys.modules access bug are unrelated." }, { "left": "issue:43531", "right": "issue:45362", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE vs Qwen3.5 chat crash; same model family but not the same concrete issue." + "reason": "Qwen3-MoE sliding_window behavior and Qwen3.5-35B chat crash are not the same issue." }, { "left": "issue:36246", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but for different packages and symbols; no shared bug." + "reason": "Different missing-import regressions: Qwen2.5 VL image processor vs is_offline_mode from huggingface_hub." }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Different missing exports from different transformer components; not the same issue." + "reason": "Both are import errors, but they refer to different missing symbols and code paths." }, { "left": "issue:43673", "right": "issue:44561", "accept": false, - "reason": "Chunked-prefill cache regression vs removal of is_torch_fx_available breaking trust_remote_code; distinct regressions." + "reason": "Chunked-prefill cache issue and removal of is_torch_fx_available are separate v5 regressions." }, { "left": "issue:44336", "right": "issue:44485", "accept": false, - "reason": "ANSI logging output bug vs GLM-5 RoPE discussion; unrelated." + "reason": "Loading-report ANSI leakage and GLM-5 RoPE implementation discussion are unrelated." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "ESM2 breakage vs Gemma 4 mm_token_type_ids defaulting issue; unrelated model families and failures." + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids defaulting are different model-specific bugs." }, { "left": "issue:44246", "right": "issue:44871", "accept": false, - "reason": "Slow import vs Gemma-3 eos_token_id mismatch; different bug classes." - }, - { - "left": "issue:44242", - "right": "issue:44857", - "accept": false, - "reason": "Router-logits load balancing loss issue vs LwDetrImageLoss AMP crash; unrelated." + "reason": "Intermittent import slowness and Gemma-3 eos_token_id inconsistency are unrelated." } ] }, @@ -9608,10 +9642,10 @@ "issue:43653", "issue:43673", "issue:43901", - "issue:43931", "issue:44060", "issue:44077", "issue:44162", + "issue:44242", "issue:44246", "issue:44295", "issue:44336", @@ -9621,7 +9655,7 @@ "issue:44734", "issue:44749", "issue:44779", - "issue:44869", + "issue:44857", "issue:44877", "issue:44933", "issue:44977", @@ -9633,12 +9667,12 @@ "issue:45229", "issue:45292", "issue:45335", - "issue:45356", "issue:45375", "issue:45459", "issue:45563" ], "soft_pairs": [ + "issue:44242|issue:44857", "issue:44779|issue:45292", "issue:44295|issue:44749", "issue:44877|issue:45375", @@ -9647,8 +9681,8 @@ "issue:36246|issue:43596", "issue:44987|issue:45459", "issue:42886|issue:45229", - "issue:44995|issue:45200", "issue:43901|issue:45003", + "issue:44995|issue:45200", "issue:41628|issue:41950", "issue:43257|issue:45137", "issue:43065|issue:43408", @@ -9657,220 +9691,205 @@ "issue:44496|issue:45003", "issue:43262|issue:44246", "issue:44060|issue:44077", - "issue:43475|issue:43540", "issue:44466|issue:45335", + "issue:43475|issue:43540", "issue:43653|issue:44162", "issue:44368|issue:44977", "issue:44368|issue:45563", "issue:44933|issue:45459", "issue:43441|issue:43606", - "issue:43673|issue:44734", - "issue:44869|issue:45356", - "issue:43931|issue:44368", - "issue:43232|issue:43673" + "issue:43673|issue:44734" ], - "prepared_review_unit_hash": "f0b2f11d85ab19d23cab8ac13ec97c071118afdf6ed9a4a4688724bdc9b4a25d", + "prepared_review_unit_hash": "7f116e26c142e9c7618f40136b634088718bc4f51c3856f74a0342a579bdcb5e", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36567, - "estimated_input_tokens": 9142, - "estimated_eval_tokens": 18540 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35291, + "estimated_input_tokens": 8823, + "estimated_eval_tokens": 17902 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items are mostly heterogeneous transformer regressions and model-specific bugs; the soft pairs are mostly broad lexical matches, not the same concrete issue. I would not merge any of the proposed soft edges.", - "confidence": 0.84, - "canonical_issue_reason": "issue:45292 is the cleanest representative anchor: it describes a concrete, broadly applicable `resize_token_embeddings` bug and is open with inbound references.", + "summary": "The proposed pairs are mostly only loosely related by subsystem or symptom. They do not look like the same concrete bug/change, so none are safe duplicate merges.", + "confidence": 0.91, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Among this mixed set, issue:45292 is the most reusable canonical bucket for related embedding-resize failures.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Different problems: MoE load-balancing loss vs an AMP/CUDA crash in image loss code." + }, { "left": "issue:44779", "right": "issue:45292", "accept": false, - "reason": "Both mention tokenizer/embedding regressions in v5, but they are different bugs: Deepseek decoding wrong results vs `resize_token_embeddings` not updating outputs." + "reason": "Both touch tokenization/model embeddings, but one is a Deepseek tokenizer regression and the other is resize_token_embeddings not propagating to output embeddings." }, { "left": "issue:44295", "right": "issue:44749", "accept": false, - "reason": "Different symptoms and code paths: a `position_ids` buffer read error vs a generic slowdown after upgrading Transformers." + "reason": "Unrelated symptoms: a position_ids buffer issue vs a performance regression after upgrading Transformers." }, { "left": "issue:44877", "right": "issue:45375", "accept": false, - "reason": "Both involve config-loading strictness, but they are model-specific and describe different failures: blocked loading vs a silently dropped field." + "reason": "Both mention config/strict handling, but they involve different models and different failure modes; not clearly the same bug." }, { "left": "issue:43441", "right": "issue:43540", "accept": false, - "reason": "Unrelated model failures: FlashAttention on Ministral-3 vs video-input validation in Qwen3OmniMoe." + "reason": "Different model paths and failures: FlashAttention on Ministral-3 vs video input processing in Qwen3OmniMoe." }, { "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "One is a tokenizer decoder-type warning in v5; the other is a Deepseek tokenizer correctness regression. Same broad area, different bug." + "reason": "Both are tokenizer-related, but one is a decoder-type warning in v5 and the other is an incorrect Deepseek tokenizer result regression." }, { "left": "issue:36246", "right": "issue:43596", "accept": false, - "reason": "Different failures: missing Qwen2.5 VL image processor import vs a DeepSpeed ZeRO3/BertModel indexing crash." + "reason": "ImportError for a specific Qwen image processor is unrelated to a DeepSpeed ZeRO3 BertModel index error." }, { "left": "issue:44987", "right": "issue:45459", "accept": false, - "reason": "Both touch loading behavior, but one is a model-loading failure on `physical-intelligence/fast`; the other is tokenizer error masking when protobuf is absent." + "reason": "Different concerns: model loading failure for a specific repo vs protobuf-related tokenizer error masking." }, { "left": "issue:42886", "right": "issue:45229", "accept": false, - "reason": "Tokenizer offline-cache loading bug is unrelated to Gemma4 multi-GPU CUDA OOM." + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are unrelated issues." }, { - "left": "issue:44995", - "right": "issue:45200", + "left": "issue:43901", + "right": "issue:45003", "accept": false, - "reason": "Stale indexer cache on second forward pass is unrelated to `mm_token_type_ids` being required for text-only fine-tuning." + "reason": "Docs mismatch for a pipeline option vs unsafe sys.modules access in modeling_utils are unrelated." }, { - "left": "issue:43901", - "right": "issue:45003", + "left": "issue:44995", + "right": "issue:45200", "accept": false, - "reason": "Docs mismatch for `TextClassificationPipeline` is unrelated to unsafe `sys.modules` access in `modeling_utils`." + "reason": "Different model-specific bugs: stale indexer cache on second forward pass vs mm_token_type_ids defaulting for Gemma 4 fine-tuning." }, { "left": "issue:41628", "right": "issue:41950", "accept": false, - "reason": "Generic `AutoImageProcessor` import failure is not the same bug as the video-classification pipeline selecting image processors." + "reason": "ImportError for AutoImageProcessor and a video-classification pipeline looking for image processors are related only broadly, not the same concrete bug." }, { "left": "issue:43257", "right": "issue:45137", "accept": false, - "reason": "Qwen3 MOE weight conversion under accelerate+deepspeed and a ZeRO3 deque pop error are different loading/runtime issues." + "reason": "Both involve DeepSpeed/ZeRO3 contexts, but one is weight conversion with accelerate and the other is an empty-deque runtime error." }, { "left": "issue:43065", "right": "issue:43408", "accept": false, - "reason": "Both are SAM3-related, but one is a dummy Conv2d concern and the other is a model-type mismatch warning; not the same defect." + "reason": "Both are SAM3-related, but one is a dummy Conv2d implementation and the other is a model-type mismatch warning." }, { "left": "issue:43295", "right": "issue:45292", "accept": false, - "reason": "Both mention embedding/tokenizer interfaces, but one is a processor.tokenizer regression and the other is output embeddings not resizing." + "reason": "Both concern embedding/tokenizer interactions, but one is a regression around processor.tokenizer and images-to-tokenizer behavior, not resize_token_embeddings." }, { "left": "issue:43232", "right": "issue:44336", "accept": false, - "reason": "Generation kwargs after sync_gpus and ANSI output from loading_report are unrelated." + "reason": "Generation kwargs handling and ANSI codes in loading_report are unrelated." }, { "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "Unrecognized model/config issue and unsafe `sys.modules` access are different loading failures." + "reason": "Unrecognized model/config loading and unsafe sys.modules access are different failures." }, { "left": "issue:43262", "right": "issue:44246", "accept": false, - "reason": "Audio chat-template sampling-rate bug is unrelated to occasional slow `import transformers` behavior." + "reason": "Audio chat-template sampling-rate defaults are unrelated to import latency." }, { "left": "issue:44060", "right": "issue:44077", "accept": false, - "reason": "Both are warnings in model config/model classes, but one is a tied-weights bug and the other is `patchtsmixer` post_init allowance." + "reason": "A tied-weights warning bug and a patchtsmixer post_init policy issue do not match." }, { - "left": "issue:43475", - "right": "issue:43540", + "left": "issue:44466", + "right": "issue:45335", "accept": false, - "reason": "Different SAM3/Qwen3OmniMoe failures with no shared concrete code path." + "reason": "Both involve weights/embeddings, but one is serialization depending on device and the other is resize_token_embeddings not updating decoder.embed_tokens." }, { - "left": "issue:44466", - "right": "issue:45335", + "left": "issue:43475", + "right": "issue:43540", "accept": false, - "reason": "Both involve tied/embedding weights, but one is inconsistent serialization by device and the other is resize propagation to decoder embeddings." + "reason": "Different SAM3 video encoder attribute bug vs Qwen3OmniMoe video-processing ValueError." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "BigBirdTokenizer special-token decoding and ESM2 breakage are separate tokenizer/model issues." + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are not the same issue." }, { "left": "issue:44368", "right": "issue:44977", "accept": false, - "reason": "Warning about `tie_word_embeddings=False` is not the same as a Qwen3.5 flash-attention generation failure." + "reason": "A tie_word_embeddings warning in Qwen3.5 fine-tuning is not the same as a flash-attention generation bug." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "A tie-weights warning and a stale `num_return_sequences` warning in paged generate are unrelated." + "reason": "Warning about tie_word_embeddings is unrelated to a stale num_return_sequences warning in paged generate." }, { "left": "issue:44933", "right": "issue:45459", "accept": false, - "reason": "Missing import from `image_utils` is unrelated to tokenizer errors hidden by protobuf import handling." + "reason": "Image_utils import problem and protobuf error masking are unrelated." }, { "left": "issue:43441", "right": "issue:43606", "accept": false, - "reason": "Different device/model issues: FlashAttention on Ministral-3 vs CPU offload device mismatch for bark-small." + "reason": "Different code paths: FlashAttention failure vs device mismatch under CPU offload." }, { "left": "issue:43673", "right": "issue:44734", "accept": false, - "reason": "Both are generation/cache-related, but one is missing cache in chunked prefill and the other is wrong indexing in `transformers serve` KV continuation." - }, - { - "left": "issue:44869", - "right": "issue:45356", - "accept": false, - "reason": "Whisper word-timestamp decode crash and Kimi-K2.5 tokenizer codec regression are distinct decoder/tokenizer bugs." - }, - { - "left": "issue:43931", - "right": "issue:44368", - "accept": false, - "reason": "Model loading/config mismatch for Qwen3-VL-30B-A3B-Instruct is unrelated to a tied-weights warning on Qwen3.5 fine-tuning." - }, - { - "left": "issue:43232", - "right": "issue:43673", - "accept": false, - "reason": "GenerationMixin kwargs handling after sync_gpus and missing cache during chunked prefill are different generation bugs." + "reason": "Both involve generation/cache, but one is missing cache in chunked prefill and the other is wrong indexing in serve continuation; not clearly the same bug." } ] }, @@ -9906,7 +9925,6 @@ "issue:43381", "issue:43475", "issue:43482", - "issue:43504", "issue:43525", "issue:43643", "issue:43646", @@ -9914,17 +9932,18 @@ "issue:43701", "issue:43704", "issue:43873", + "issue:43931", "issue:44038", "issue:44162", "issue:44188", "issue:44222", "issue:44297", - "issue:44351", "issue:44361", - "issue:44496", + "issue:44368", "issue:44623", "issue:44661", "issue:44749", + "issue:44869", "issue:44933", "issue:44945", "issue:45081", @@ -9936,29 +9955,32 @@ "issue:45292", "issue:45310", "issue:45335", + "issue:45356", "issue:45362", "issue:45375", "issue:45381", "issue:45397", - "issue:45478", - "issue:45561" + "issue:45478" ], "soft_pairs": [ + "issue:44869|issue:45356", + "issue:43931|issue:44368", + "issue:43232|issue:43673", "issue:44297|issue:44623", "issue:43673|issue:45278", "issue:43381|issue:45292", "issue:43646|issue:44162", "issue:43381|issue:45335", + "issue:44361|issue:44749", "issue:28282|issue:30990", "issue:42673|issue:43704", - "issue:44361|issue:44749", "issue:44749|issue:45310", "issue:45230|issue:45397", "issue:44038|issue:44623", - "issue:43475|issue:45381", "issue:43475|issue:43643", "issue:42994|issue:43482", "issue:43701|issue:44749", + "issue:43475|issue:45381", "issue:42371|issue:45092", "issue:45081|issue:45103", "issue:45375|issue:45478", @@ -9966,243 +9988,240 @@ "issue:45310|issue:45375", "issue:42222|issue:45200", "issue:43873|issue:44945", - "issue:43475|issue:44661", "issue:42994|issue:43525", - "issue:45362|issue:45375", + "issue:43475|issue:44661", "issue:42831|issue:44222", + "issue:45362|issue:45375", "issue:45310|issue:45397", "issue:43232|issue:44188", - "issue:44222|issue:45310", - "issue:41628|issue:44351", - "issue:43504|issue:45561", - "issue:44496|issue:45561" + "issue:44222|issue:45310" ], - "prepared_review_unit_hash": "ed0b7c39d1a0740677accc1333c74cf64036aaf96e425bded929afa9e734db99", + "prepared_review_unit_hash": "fd539651e5222e33ea58948b9935f39f51f9c32a6b6015f48e02a33864d308ba", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 36842, - "estimated_input_tokens": 9211, - "estimated_eval_tokens": 18678 + "serialized_chars": 36957, + "estimated_input_tokens": 9240, + "estimated_eval_tokens": 18736 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose mix of unrelated Transformers v5 regressions, model-specific load/config failures, and a few import/error reports. Only one soft edge looks like a true duplicate pair; most others share a broad subsystem but not the same underlying bug. The best representative issue is the Qwen3.5 MoE from_pretrained regression, which has a clear duplicate counterpart and a concrete failure mode.", - "confidence": 0.78, - "canonical_issue_reason": "issue:45310 is the cleanest concrete duplicate-ready report here: it describes a specific Qwen3.5 MoE from_pretrained failure, and it has an exact duplicate counterpart (issue:45478).", + "summary": "This cluster is mostly unrelated issue noise, but it contains one clear duplicate pair about the same Qwen3ForCausalLM VRAM leak in multi-threaded dataloader use. No PRs are present.", + "confidence": 0.64, + "canonical_issue_reason": "issue:42673 is the strongest canonical issue: it is specific, concrete, and has an exact duplicate in issue:43704 with the same title and bug description.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the strongest representative issue because it is specific, recurring, and clearly actionable, unlike the many broader or model-agnostic regressions in the set.", + "best_issue_reason": "issue:42673 is the best representative issue overall because it describes a clear, actionable bug and matches an exact duplicate; the rest of the cluster appears thematically mixed.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Different tokenizer/runtime failures: Whisper trailing replacement-character crash vs Kimi-K2.5 codec/fix_mistral_regex regression." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Different problems and models: Qwen3-VL weight-shape mismatch vs a Qwen3.5 warning about tie_word_embeddings during LoRA fine-tuning." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both are generation-related, but one is after sync_gpus and the other is missing cache during chunked prefill; not the same code-path bug." + }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "Different bugs: tokenizer_class metadata mismatch vs processor.save_pretrained missing files." + "reason": "Both involve save_pretrained, but one is tokenizer metadata and the other is missing processor files; different artifacts and failures." }, { "left": "issue:43673", "right": "issue:45278", "accept": false, - "reason": "Both are version-related regressions, but one is a generation cache bug and the other is a broad import-error report." + "reason": "A specific cache bug in chunked prefill is not the same as broad import errors after upgrading versions." }, { "left": "issue:43381", "right": "issue:45292", "accept": false, - "reason": "Unrelated code paths: gradient-checkpointing eval-mode behavior vs resize_token_embeddings/output_embeddings sync." + "reason": "Gradient checkpointing eval-mode restriction is unrelated to resize_token_embeddings not updating output embeddings." }, { "left": "issue:43646", "right": "issue:44162", "accept": false, - "reason": "Both involve model compatibility, but one is trust_remote_code missing fields and the other is an ESM2 breakage." + "reason": "Missing fields from trust_remote_code AutoConfig is unrelated to ESM2 being broken." }, { "left": "issue:43381", "right": "issue:45335", "accept": false, - "reason": "Different bugs despite both touching training/model internals; no shared underlying failure." + "reason": "Eval-mode gradient checkpointing issue is unrelated to t5gemma embedding resize not propagating." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "MLukeTokenizer AttributeError is unrelated to a Chinese-language performance regression." }, { "left": "issue:28282", "right": "issue:30990", "accept": false, - "reason": "ImportError for missing PyTorch is not the same as Sentence Transformers hanging during load." + "reason": "PyTorch import-missing error is not the same as Sentence Transformers getting stuck while loading." }, { "left": "issue:42673", "right": "issue:43704", "accept": true, - "reason": "Same issue: the Qwen3ForCausalLM VRAM leak under multiple dataloader threads is described identically." - }, - { - "left": "issue:44361", - "right": "issue:44749", - "accept": false, - "reason": "Tokenizer AttributeError and performance slowdown are unrelated failures." + "reason": "Exact same title and bug: Qwen3ForCausalLM VRAM leak when used in multiple dataloader threads." }, { "left": "issue:44749", "right": "issue:45310", "accept": false, - "reason": "Chinese filtering slowdown is unrelated to the Qwen3.5 MoE from_pretrained crash." + "reason": "Performance slowdown from upgrading is unrelated to Qwen3.5 Moe from_pretrained failure." }, { "left": "issue:45230", "right": "issue:45397", "accept": false, - "reason": "The generic bug report is too vague and does not identify the same Gemma 4 zero3 loading failure." + "reason": "Generic bug report vs a specific Gemma-4 zero3 from_pretrained problem; no concrete overlap." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Qwen3-VL-Moe breakage and processor.save_pretrained missing files are different problems." - }, - { - "left": "issue:43475", - "right": "issue:45381", - "accept": false, - "reason": "Both are vision/video model issues, but they affect different models and different fields/paths." + "reason": "Qwen3-VL-Moe v5 bug is unrelated to processor.save_pretrained missing files." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "SAM 3 video output attribute error is unrelated to trust_remote_code missing fields." + "reason": "SAM 3 VisionEncoder missing attribute is unrelated to AutoConfig returning missing fields under trust_remote_code." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized saving failure and GGUF loading failure are distinct quantization-related bugs." + "reason": "Quantized model saving failure is unrelated to Qwen2.5-GGUF loading failure." }, { "left": "issue:43701", "right": "issue:44749", "accept": false, - "reason": "Checkpoint resume key mismatch is unrelated to the filtering slowdown report." + "reason": "Checkpoint resume key mismatch is unrelated to the data filtering slowdown regression." + }, + { + "left": "issue:43475", + "right": "issue:45381", + "accept": false, + "reason": "Different multimodal bugs: SAM 3 missing attribute vs qwen2.5-vl video position IDs being wrong." }, { "left": "issue:42371", "right": "issue:45092", "accept": false, - "reason": "TF32 settings behavior has nothing to do with InternVL2 meta-initialization incompatibility." + "reason": "TF32 API guidance issue is unrelated to InternVL2 remote-code/meta-init incompatibility." }, { "left": "issue:45081", "right": "issue:45103", "accept": false, - "reason": "Tokenizer regex patch crash and auto-docstring annotation crash are unrelated." + "reason": "Both are AttributeErrors, but one is Mistral tokenizer regex patching and the other is auto_docstring annotations handling." }, { "left": "issue:45375", "right": "issue:45478", "accept": false, - "reason": "Same model family, but one is a missing config field in vision config and the other is a generic from_pretrained error; not enough evidence of the same bug." + "reason": "Related Qwen3.5 Moe family, but one is a missing config field in vision config and the other is a from_pretrained error; not enough to merge." }, { "left": "issue:41628", "right": "issue:44933", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols/modules and different root causes." + "reason": "Missing AutoImageProcessor import and nonexistent image_utils import are different import-registry issues." }, { "left": "issue:45310", "right": "issue:45375", "accept": false, - "reason": "Qwen3.5 MoE load failure and Qwen3_5MoeVisionConfig missing field are related at most by model family, not the same bug." + "reason": "Both concern Qwen3.5 Moe, but one is a from_pretrained error and the other is a dropped vision config field; not the same bug." }, { "left": "issue:42222", "right": "issue:45200", "accept": false, - "reason": "Vitpose model breakage and Gemma 4 token-type defaults are unrelated." + "reason": "VitPose model breakage is unrelated to Gemma 4 mm_token_type_ids defaulting." }, { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Quantization offloading issues and pipeline-parallel output errors are different execution paths." + "reason": "Quantization offloading issue is unrelated to incorrect outputs under pipeline parallelism." }, { - "left": "issue:43475", - "right": "issue:44661", + "left": "issue:42994", + "right": "issue:43525", "accept": false, - "reason": "SAM 3 output attribute issue is unrelated to tokenizer mapping handling." + "reason": "Quantized save failure is unrelated to Llama4Config missing pad_token_id." }, { - "left": "issue:42994", - "right": "issue:43525", + "left": "issue:43475", + "right": "issue:44661", "accept": false, - "reason": "Quantized save failure and missing pad_token_id in Llama4Config are unrelated." + "reason": "SAM 3 output attribute error is unrelated to add-new-model-like failing for tokenizer-mapped models." }, { - "left": "issue:45362", - "right": "issue:45375", + "left": "issue:42831", + "right": "issue:44222", "accept": false, - "reason": "Both involve Qwen3.5-related reports, but one is a chat crash and the other is a missing vision config field." + "reason": "FineGrainedFP8 accuracy regression is not the same as FP8 save_pretrained failure for moe models." }, { - "left": "issue:42831", - "right": "issue:44222", + "left": "issue:45362", + "right": "issue:45375", "accept": false, - "reason": "FP8 accuracy degradation and FP8 save_pretrained failure are different bugs." + "reason": "Qwen3.5-35B chat crash is unrelated to a missing deepstack_visual_indexes field in a different config class." }, { "left": "issue:45310", "right": "issue:45397", "accept": false, - "reason": "Qwen3.5 MoE from_pretrained error is unrelated to Gemma 4 zero3 loading." + "reason": "Qwen3.5 Moe from_pretrained error is unrelated to Gemma-4 zero3 from_pretrained behavior." }, { "left": "issue:43232", "right": "issue:44188", "accept": false, - "reason": "Generation kwargs/sync_gpus behavior and torch.compile attention-kernel divergence are different issues." + "reason": "Both mention generation, but one is a sync_gpus kwargs update bug and the other is divergent attention kernels under torch.compile." }, { "left": "issue:44222", "right": "issue:45310", "accept": false, - "reason": "FP8 save_pretrained for moe models is unrelated to the Qwen3.5 MoE from_pretrained crash." - }, - { - "left": "issue:41628", - "right": "issue:44351", - "accept": false, - "reason": "Cannot import AutoImageProcessor and cannot import HybridCache are separate import regressions." - }, - { - "left": "issue:43504", - "right": "issue:45561", - "accept": false, - "reason": "Legacy-field BEiT loading failure and pytest-xdist race in testing utils are unrelated." - }, - { - "left": "issue:44496", - "right": "issue:45561", - "accept": false, - "reason": "Unrecognized model config and captured_info.txt race are completely different problems." + "reason": "FP8 save_pretrained for moe models is unrelated to Qwen3.5 Moe from_pretrained loading error." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: issue:45310 and issue:45478 are clearly the same Qwen3.5 MoE from_pretrained bug by title, and the rest of the soft-edge calls stay conservative by treating only clearly identical or obviously unrelated cases as duplicates/non-duplicates. I don\u2019t see evidence of overstating any pairings." + "feedback": "The summary is grounded in the packet: it correctly identifies a single exact duplicate pair (issue:42673 / issue:43704) and appropriately characterizes the rest as mostly unrelated. The canonical/best-issue rationale is conservative, and there are no PRs in the packet. The soft-edge rejections are also cautious and do not overclaim equivalence." }, "evaluator_used": true, "retried": false, @@ -10232,9 +10251,9 @@ "issue:42994", "issue:43232", "issue:43278", - "issue:43295", "issue:43377", "issue:43479", + "issue:43504", "issue:43526", "issue:43673", "issue:43723", @@ -10248,12 +10267,11 @@ "issue:44079", "issue:44164", "issue:44188", - "issue:44222", "issue:44242", - "issue:44297", + "issue:44351", "issue:44373", - "issue:44442", "issue:44448", + "issue:44496", "issue:44568", "issue:44623", "issue:44734", @@ -10273,8 +10291,11 @@ "issue:45561" ], "soft_pairs": [ - "issue:43526|issue:45561", + "issue:41628|issue:44351", + "issue:43504|issue:45561", + "issue:44496|issue:45561", "issue:42994|issue:44164", + "issue:43526|issue:45561", "issue:43232|issue:44792", "issue:44038|issue:45310", "issue:43377|issue:44242", @@ -10283,8 +10304,8 @@ "issue:36010|issue:41628", "issue:43479|issue:44079", "issue:42898|issue:44568", - "issue:43479|issue:44869", "issue:44964|issue:45137", + "issue:43479|issue:44869", "issue:43976|issue:44933", "issue:42175|issue:44933", "issue:43784|issue:43824", @@ -10294,231 +10315,228 @@ "issue:44060|issue:45125", "issue:44734|issue:45103", "issue:42898|issue:45356", - "issue:43723|issue:44568", "issue:45137|issue:45276", + "issue:43723|issue:44568", "issue:42898|issue:44779", "issue:43874|issue:45092", "issue:44077|issue:44242", "issue:42898|issue:44448", - "issue:41628|issue:45459", - "issue:44222|issue:44297", - "issue:42898|issue:43295", - "issue:44442|issue:45459" + "issue:41628|issue:45459" ], - "prepared_review_unit_hash": "a37c040804e12a57a8472ae7544a5670dbfff9581eab60a351a91e8783fa33af", + "prepared_review_unit_hash": "7ddd3724fee4d8ab7da37e644fd8cab6b22a2bb9013b329f40723a7716aa0c6c", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 31, - "serialized_chars": 37004, - "estimated_input_tokens": 9251, - "estimated_eval_tokens": 18758 + "serialized_chars": 36481, + "estimated_input_tokens": 9121, + "estimated_eval_tokens": 18498 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: many items are superficially similar (tokenizer/load/save/import regressions) but point to different concrete failures. I would not merge any of the suggested soft pairs as duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "No single canonical issue; the set is too mixed, with only a loose tokenizer-regression subgroup and many unrelated model/load/save bugs.", + "summary": "The cluster is highly heterogeneous: most items are distinct regression reports across unrelated model families, tokenizer paths, config/loading code, and test/infrastructure issues. No soft edge looks like a true duplicate pair.", + "confidence": 0.18, + "canonical_issue_reason": "42898 is the broadest, most representative issue in the set because it frames a v5 behavior regression with external reference activity, but the cluster overall does not form a cohesive duplicate group.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42898 is the broadest representative of the tokenizer-regression subgroup (v5 behavior change) and has the most general framing, though it still does not cover the unrelated items.", + "best_issue_reason": "42898 is the best available anchor for triage, though it still only represents one specific tokenizer-behavior regression and not the rest of the unrelated issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43526", + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols and likely different API breakages." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "Semantic segmentation preset loading and a pytest-xdist file race are unrelated." + }, + { + "left": "issue:44496", "right": "issue:45561", "accept": false, - "reason": "Different subsystems: BeitImageProcessorFast label reduction vs pytest-xdist race in testing utils." + "reason": "Unrecognized model config loading is unrelated to a testing-utils race condition." }, { "left": "issue:42994", "right": "issue:44164", "accept": false, - "reason": "Both involve saving/loading, but one is quantized model saving and the other is extra_state handling; different bugs." + "reason": "Both concern saving/loading, but one is quantized model saving and the other is extra_state serialization handling." + }, + { + "left": "issue:43526", + "right": "issue:45561", + "accept": false, + "reason": "A BEiT image processor label-reduction bug is unrelated to a captured_info.txt race." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Both touch generation, but the failures are unrelated: KV-arg sync handling vs janus image-generation test." + "reason": "Generation kwargs handling and a Janus image-generation test failure do not match the same underlying bug." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Same model family, but different concrete issues and versions: Qwen3-VL-Moe vs Qwen3.5 MoE from_pretrained." + "reason": "Both mention Qwen/Moe, but they appear to be different model/version-specific from_pretrained failures." }, { "left": "issue:43377", "right": "issue:44242", "accept": false, - "reason": "Unrelated bugs: MIMI padding-mask divergence vs MoE load-balancing loss omission." + "reason": "Missing padding-mask support in MIMI encoder is unrelated to load-balancing loss wiring." }, { "left": "issue:30990", "right": "issue:35141", "accept": false, - "reason": "Sentence Transformers loading hang is unrelated to token-embedding resizing reinit in post_init." + "reason": "Sentence Transformers loading hangs and embedding reinitialization after resize are different problems." }, { "left": "issue:43673", "right": "issue:44188", "accept": false, - "reason": "Both are generation-related, but one is missing cache in chunked_prefill and the other is torch.compile branching divergence." + "reason": "GenerationMixin cache issues and attention-kernel divergence under torch.compile are separate code paths." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but for different symbols and paths: GenerationMixin vs AutoImageProcessor." + "reason": "Both are import errors, but they concern different classes and different failing imports." }, { "left": "issue:43479", "right": "issue:44079", "accept": false, - "reason": "Both mention None-handling, but one is multimodal config defaults and the other is ModelOutput key assignment." + "reason": "Config default initialization and ModelOutput key assignment are different object-model bugs." }, { "left": "issue:42898", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer regressions, but different behaviors and code paths: clean_up_tokenization_spaces vs add_special_tokens BOS/EOS." + "reason": "Both are tokenizer regressions in v5, but they affect different behaviors and different tokenizer code paths." }, { - "left": "issue:43479", - "right": "issue:44869", + "left": "issue:44964", + "right": "issue:45137", "accept": false, - "reason": "Unrelated: multimodal config initialization vs Whisper word-timestamp decoding crash." + "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque underflow are unrelated." }, { - "left": "issue:44964", - "right": "issue:45137", + "left": "issue:43479", + "right": "issue:44869", "accept": false, - "reason": "Different failure classes: Phi-4 multimodal loading vs DeepSpeed ZeRO3 deque underflow." + "reason": "Multimodal config defaults and Whisper timestamp decoding are unrelated." }, { "left": "issue:43976", "right": "issue:44933", "accept": false, - "reason": "Python-version support issue is unrelated to a missing import from image_utils." + "reason": "Python-version compatibility and a missing image_utils import are unrelated." }, { "left": "issue:42175", "right": "issue:44933", "accept": false, - "reason": "Package extra/backend selection is unrelated to the image_utils import regression." + "reason": "Packaging backend extras and a missing import symbol are unrelated." }, { "left": "issue:43784", "right": "issue:43824", "accept": false, - "reason": "Both are import-time breakages, but one is sentence-transformers nn import and the other is a missing Qwen2.5 VL class." + "reason": "These are different import failures in different packages/symbols." }, { "left": "issue:43278", "right": "issue:45137", "accept": false, - "reason": "Embedding dtype drift in eval is unrelated to the ZeRO3 deque IndexError." + "reason": "Embedding dtype drift during eval is unrelated to DeepSpeed ZeRO3 deque errors." }, { "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "Processor save_pretrained missing files is unrelated to Qwen3.5 MoE from_pretrained failure." + "reason": "Processor save-pretrained file omissions and Qwen3.5 MoE from_pretrained errors are different issues." }, { "left": "issue:44188", "right": "issue:44373", "accept": false, - "reason": "torch.compile attention-kernel divergence is unrelated to a wrong docstring." + "reason": "Torch.compile attention-kernel divergence and a docstring issue are unrelated." }, { "left": "issue:44060", "right": "issue:45125", "accept": false, - "reason": "Both are Qwen-related, but one is a tied-weights warning bug and the other is missing _tp_plan for tensor parallelism." + "reason": "A tied-weights warning bug and missing tensor-parallel plan metadata are different model-integration issues." }, { "left": "issue:44734", "right": "issue:45103", "accept": false, - "reason": "KV cache continuation crash in serve is unrelated to auto_docstring annotation handling." + "reason": "KV-cache continuation indexing and auto-docstring annotation handling are unrelated." }, { "left": "issue:42898", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer regressions in v5, but they affect different models and different root causes." + "reason": "Both are tokenizer regressions, but they concern different models and different regression causes." }, { - "left": "issue:43723", - "right": "issue:44568", + "left": "issue:45137", + "right": "issue:45276", "accept": false, - "reason": "Tokenizer loading failure and BOS/EOS insertion regression are different bugs." + "reason": "DeepSpeed deque underflow and Gemma4 resize-token-embeddings behavior are unrelated." }, { - "left": "issue:45137", - "right": "issue:45276", + "left": "issue:43723", + "right": "issue:44568", "accept": false, - "reason": "DeepSpeed ZeRO3 deque issue is unrelated to Gemma4 resize_token_embeddings not updating tied embeddings." + "reason": "Tokenizer loading failure and missing BOS/EOS insertion are different tokenizer issues." }, { "left": "issue:42898", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but the affected tokenizer and failure mode differ." + "reason": "Both are v5 tokenizer regressions, but they affect different tokenizers and different behavior." }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Missing image-patch method in GLM46V is unrelated to InternVL2 meta-init incompatibility." + "reason": "Missing multimodal helper method and remote-code/meta-init incompatibility are distinct multimodal problems." }, { "left": "issue:44077", "right": "issue:44242", "accept": false, - "reason": "Optional post_init gating for patchtsmixer is unrelated to missing load-balancing loss." + "reason": "Optional post_init handling and load-balancing loss computation are not the same bug." }, { "left": "issue:42898", "right": "issue:44448", "accept": false, - "reason": "Both are v5 output regressions, but one is token cleanup behavior and the other is Pegasus generation output drift." + "reason": "Tokenizer cleanup behavior changes and Pegasus output drift are different regression reports." }, { "left": "issue:41628", "right": "issue:45459", "accept": false, - "reason": "ImportError for AutoImageProcessor is unrelated to protobuf error masking in tokenizer loading." - }, - { - "left": "issue:44222", - "right": "issue:44297", - "accept": false, - "reason": "Both are save_pretrained-related, but one is FP8 MoE model saving and the other is tokenizer metadata mismatch." - }, - { - "left": "issue:42898", - "right": "issue:43295", - "accept": false, - "reason": "Both are regressions, but one is tokenizer cleanup behavior and the other is processor.tokenizer/image-passing API breakage." - }, - { - "left": "issue:44442", - "right": "issue:45459", - "accept": false, - "reason": "FastSpeech2Conformer tokenizer loading and protobuf error masking are different issues." + "reason": "Missing top-level import and tokenizer protobuf-error masking are different import/tokenizer issues." } ] }, @@ -10545,16 +10563,16 @@ "issue:22355", "issue:28282", "issue:30990", - "issue:33357", "issue:38175", "issue:41628", - "issue:42371", "issue:42757", "issue:42831", "issue:42886", + "issue:42898", "issue:42994", "issue:43066", "issue:43122", + "issue:43295", "issue:43479", "issue:43576", "issue:43644", @@ -10586,29 +10604,32 @@ "issue:45092", "issue:45103", "issue:45278", - "issue:45290", "issue:45310", "issue:45357", - "issue:45406" + "issue:45406", + "issue:45459" ], "soft_pairs": [ + "issue:44222|issue:44297", + "issue:42898|issue:43295", + "issue:44442|issue:45459", "issue:43479|issue:43644", "issue:44222|issue:44623", "issue:42831|issue:44623", "issue:44206|issue:45103", "issue:22355|issue:28282", "issue:44297|issue:45310", - "issue:44483|issue:45406", "issue:41628|issue:42757", + "issue:44483|issue:45406", "issue:44222|issue:44484", "issue:44263|issue:44360", "issue:43723|issue:44977", "issue:44263|issue:45092", - "issue:44295|issue:44623", "issue:44514|issue:45103", + "issue:44295|issue:44623", "issue:43479|issue:44913", - "issue:43723|issue:44871", "issue:44373|issue:44485", + "issue:43723|issue:44871", "issue:44261|issue:45092", "issue:43723|issue:44991", "issue:43723|issue:44442", @@ -10620,260 +10641,266 @@ "issue:42831|issue:44038", "issue:43576|issue:43784", "issue:43122|issue:44521", - "issue:42757|issue:45278", "issue:44263|issue:44373", + "issue:42757|issue:45278", "issue:44117|issue:44704", "issue:42886|issue:43066", - "issue:43122|issue:44514", - "issue:30990|issue:33357", - "issue:44704|issue:45290", - "issue:42371|issue:44263" + "issue:43122|issue:44514" ], - "prepared_review_unit_hash": "527d30c6554f79f3217c7169276e403e739aaf24e0327c35ad910903bc9aa380", + "prepared_review_unit_hash": "93be64134c587c9f16a8bcedef4e3239c2da9cdff8161620e7b5e5f702431282", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 36, - "serialized_chars": 38062, - "estimated_input_tokens": 9516, - "estimated_eval_tokens": 19288 + "serialized_chars": 38100, + "estimated_input_tokens": 9525, + "estimated_eval_tokens": 19306 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is highly heterogeneous: most items are unrelated regressions, import errors, or save/load edge cases. No soft-edge pair looks like a true duplicate bug/change, so I rejected them all. The best representative issue is the broad tokenizer-loading regression in v5.", - "confidence": 0.89, - "canonical_issue_reason": "issue:43723 is the broadest and most central user-facing regression here, covering tokenizer loading failures in v5 and overlapping thematically with several other tokenizer/import issues.", + "summary": "This cluster is mostly heterogeneous, but a few issues sit around a shared v5 tokenizer-loading regression theme. Only one soft pair looks like a plausible duplicate; the rest are different bugs or different code paths.", + "confidence": 0.61, + "canonical_issue_reason": "issue:43723 is the broadest representative of the tokenizer-loading regressions in v5, and it can plausibly absorb specific model-level reports in that area.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43723 is the strongest umbrella representative for the cluster\u2019s recurring loading/import regressions, even though the cluster is not actually a duplicate set.", + "best_issue_reason": "issue:43723 is the best single issue for the dominant subtheme (tokenizer loading failures), though it is not a true umbrella for the whole mixed cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "FP8/MoE save_pretrained failure vs tokenizer_class metadata mismatch; different bugs." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Tokenizer cleanup behavior change vs processor/tokenizer image-handling regression; different code paths." + }, + { + "left": "issue:44442", + "right": "issue:45459", + "accept": false, + "reason": "AutoTokenizer cannot load a specific tokenizer vs protobuf masking real tokenizer errors; not the same failure." + }, { "left": "issue:43479", "right": "issue:43644", "accept": false, - "reason": "Different bugs: multimodal config defaults vs non-persistent buffer corruption." + "reason": "Default config initialization bug vs non-persistent buffer corruption; unrelated." }, { "left": "issue:44222", "right": "issue:44623", "accept": false, - "reason": "Both mention save_pretrained, but one is FP8/MoE model saving and the other is processor file omission." + "reason": "FP8/MoE save_pretrained failure vs missing processor save files; different symptoms and causes." }, { "left": "issue:42831", "right": "issue:44623", "accept": false, - "reason": "Accuracy regression in FineGrainedFP8 is unrelated to missing processor save files." + "reason": "Accuracy regression in FP8 vs processor save_pretrained file omission; unrelated." }, { "left": "issue:44206", "right": "issue:45103", "accept": false, - "reason": "Feature extractor argument crash and auto-docstring annotation crash are unrelated code paths." + "reason": "Auto-docstring crash with feature extractor args vs AttributeError under future annotations; same area, but different concrete bugs." }, { "left": "issue:22355", "right": "issue:28282", "accept": false, - "reason": "Different import failures: missing transformers.onnx module vs missing PyTorch dependency." + "reason": "ONNX module import failure vs missing PyTorch dependency error; unrelated." }, { "left": "issue:44297", "right": "issue:45310", "accept": false, - "reason": "Tokenizer config save mismatch is not the same bug as Qwen3.5 MoE from_pretrained failure." + "reason": "Tokenizer save metadata mismatch vs Qwen3.5 MoE from_pretrained failure; different issues." }, { - "left": "issue:44483", - "right": "issue:45406", + "left": "issue:41628", + "right": "issue:42757", "accept": false, - "reason": "Chat/completions request handling and Gemma4Processor serve crash are different failures." + "reason": "Missing transformers export for AutoImageProcessor vs missing is_offline_mode import from huggingface_hub; different packages and failures." }, { - "left": "issue:41628", - "right": "issue:42757", + "left": "issue:44483", + "right": "issue:45406", "accept": false, - "reason": "Different missing symbols from different packages; same broad symptom only." + "reason": "Chat/completions request acceptance bug vs Gemma4Processor missing _tokenizer; different components." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "FP8/MoE save failure is unrelated to the max_shard_size question." + "reason": "FP8 save_pretrained bug vs max_shard_size question; not duplicates." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Both involve DSA indexer wording, but the concrete bug differs: torch.split return handling vs missing ReLU." + "reason": "torch.split return handling vs missing ReLU in DSA indexer; different concrete bugs." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Tokenizer loading regression and flash-attention generation issue are not the same underlying bug." + "reason": "Generic tokenizer loading regression vs Qwen3.5 flash-attention generation bug; different paths." }, { "left": "issue:44263", "right": "issue:45092", "accept": false, - "reason": "DSA indexer split behavior is unrelated to remote-code checkpoint meta-init incompatibility." + "reason": "DSA indexer split bug vs InternVL2 meta-init incompatibility; unrelated." }, { - "left": "issue:44295", - "right": "issue:44623", + "left": "issue:44514", + "right": "issue:45103", "accept": false, - "reason": "Reading position_ids after buffer registration is unrelated to processor save_pretrained missing files." + "reason": "Qwen2_5_VLProcessor batched chat-template crash vs auto_docstring future-annotations crash; unrelated." }, { - "left": "issue:44514", - "right": "issue:45103", + "left": "issue:44295", + "right": "issue:44623", "accept": false, - "reason": "Chat-template batching crash and annotation-introspection crash are unrelated." + "reason": "position_ids buffer read error vs processor.save_pretrained missing files; different issues." }, { "left": "issue:43479", "right": "issue:44913", "accept": false, - "reason": "Both are config-related, but they affect different models and different reload paths." + "reason": "Phi4MultimodalConfig defaulting bug vs GPTNeoXConfig rotary_pct reload bug; different configs." }, { - "left": "issue:43723", - "right": "issue:44871", + "left": "issue:44373", + "right": "issue:44485", "accept": false, - "reason": "General tokenizer loading regression vs eos_token_id config inconsistency are different bugs." + "reason": "Docstring text error vs GLM-5 RoPE implementation bug; not the same." }, { - "left": "issue:44373", - "right": "issue:44485", + "left": "issue:43723", + "right": "issue:44871", "accept": false, - "reason": "Docstring typo and RoPE implementation bug are unrelated." + "reason": "Tokenizer loading regression vs eos_token_id config inconsistency; different model issues." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "MLA rms_norm_eps precision issue is unrelated to old remote-code checkpoint meta initialization." + "reason": "Missing rms_norm_eps in MLA q_a_layernorm vs InternVL2 meta-init incompatibility; unrelated." }, { "left": "issue:43723", "right": "issue:44991", - "accept": false, - "reason": "Both are tokenizer-related, but one is a generic load regression and the other is a model-specific generation failure." + "accept": true, + "reason": "Both describe the same v5 AutoTokenizer loading regression; 44991 is a specific model instance of the broader loading failure." }, { "left": "issue:43723", "right": "issue:44442", "accept": false, - "reason": "Generic tokenizer load regression is not the same as FastSpeech2ConformerTokenizer failing to load." + "reason": "Generic tokenizer loading regression vs FastSpeech2ConformerTokenizer-specific load failure; related area but not clearly the same bug." }, { "left": "issue:30990", "right": "issue:38175", "accept": false, - "reason": "SentenceTransformers loading hang and zero-probability model outputs are unrelated." + "reason": "Sentence-Transformers loading hang vs SIGLIP2 zero-probability output; unrelated." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Remote-code meta-init incompatibility is unrelated to incorrect visual encoder keys on save_pretrained." + "reason": "Meta initialization incompatibility vs Qwen3.5 save_pretrained visual key regression; different bugs." }, { "left": "issue:42886", "right": "issue:42994", "accept": false, - "reason": "Offline tokenizer cache loading and quantized model saving are different code paths." + "reason": "Offline tokenizer cache load failure vs quantized model save failure; unrelated." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FineGrainedFP8 accuracy regression does not match the Qwen3.5 MoE from_pretrained error." + "reason": "FP8 accuracy issue vs Qwen3.5 MoE from_pretrained error; different failures." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they are different missing exports from different modules." + "reason": "Different missing imports for different classes; likely separate export issues." }, { "left": "issue:42831", "right": "issue:44038", "accept": false, - "reason": "FP8 accuracy issue and Qwen3-VL-Moe bug are not the same failure." + "reason": "FP8 accuracy regression vs Qwen3-VL-Moe bug; unrelated." }, { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "Broken env command and sentence-transformers import NameError are unrelated." + "reason": "transformers env CLI regression vs sentence-transformers import NameError; different code paths." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Tokenization changes and multimodal assistant mask generation are different regressions." + "reason": "General tokenization output changes vs multimodal assistant_mask bug; different symptoms." }, { - "left": "issue:42757", - "right": "issue:45278", + "left": "issue:44263", + "right": "issue:44373", "accept": false, - "reason": "45278 is a broad umbrella of import errors; it does not specifically match the is_offline_mode import failure." + "reason": "Indexing bug vs docstring typo; unrelated." }, { - "left": "issue:44263", - "right": "issue:44373", + "left": "issue:42757", + "right": "issue:45278", "accept": false, - "reason": "Runtime DSA indexer bug vs docstring issue are not duplicates." + "reason": "Specific missing import from huggingface_hub vs broad import-errors report; same broad area but not same concrete bug." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES returning None and cached_file kwarg forwarding are different bugs." + "reason": "TOKENIZER_MAPPING_NAMES None handling vs AutoProcessor kwargs propagation; different issues." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Offline cache loading failure and wrong tokenizer decoder type are not the same underlying issue." + "reason": "Offline cache tokenizer load failure vs wrong tokenizer decoder type in v5; different bugs." }, { "left": "issue:43122", "right": "issue:44514", "accept": false, - "reason": "Different tokenization-related regressions in different APIs and model families." - }, - { - "left": "issue:44704", - "right": "issue:45290", - "accept": false, - "reason": "Missing cached_file kwargs and assistant-message tool-call crash are unrelated." - }, - { - "left": "issue:42371", - "right": "issue:44263", - "accept": false, - "reason": "TF32 settings behavior and DSA indexer split-return bug are unrelated." + "reason": "Version-to-version tokenization difference vs batched apply_chat_template crash; different failure modes." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The packet does show a heterogeneous cluster with a tokenizer-loading subtheme, and issue:43723 is a reasonable broad representative for that subtheme. The lone accepted soft pair (43723/44991) is plausible, while the other false verdicts stay safely on the conservative side rather than overclaiming duplicates." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -10892,10 +10919,10 @@ "nodes": [ "issue:29127", "issue:30990", + "issue:33357", "issue:36296", "issue:39401", "issue:41628", - "issue:42222", "issue:42371", "issue:42757", "issue:42831", @@ -10905,10 +10932,8 @@ "issue:43232", "issue:43278", "issue:43408", - "issue:43441", "issue:43452", "issue:43475", - "issue:43784", "issue:43824", "issue:43937", "issue:43976", @@ -10916,6 +10941,7 @@ "issue:44038", "issue:44222", "issue:44261", + "issue:44263", "issue:44295", "issue:44297", "issue:44351", @@ -10936,12 +10962,15 @@ "issue:45216", "issue:45245", "issue:45278", - "issue:45446", + "issue:45290", "issue:45478" ], "soft_pairs": [ - "issue:42907|issue:44945", + "issue:30990|issue:33357", + "issue:44704|issue:45290", + "issue:42371|issue:44263", "issue:43232|issue:45245", + "issue:42907|issue:44945", "issue:44295|issue:44297", "issue:43232|issue:44743", "issue:43976|issue:44945", @@ -10950,13 +10979,13 @@ "issue:43408|issue:43475", "issue:44625|issue:44829", "issue:41628|issue:45278", - "issue:29127|issue:30990", "issue:44360|issue:44485", + "issue:29127|issue:30990", "issue:43116|issue:45200", "issue:39401|issue:44521", + "issue:44261|issue:44373", "issue:43122|issue:44779", "issue:44038|issue:44297", - "issue:44261|issue:44373", "issue:42371|issue:42831", "issue:44038|issue:44295", "issue:42907|issue:44964", @@ -10969,272 +10998,234 @@ "issue:43278|issue:44945", "issue:43824|issue:44351", "issue:42907|issue:44222", - "issue:42757|issue:44351", - "issue:42222|issue:43441", - "issue:43784|issue:45446" + "issue:42757|issue:44351" ], - "prepared_review_unit_hash": "6ce6f2e3dee04eddceab18ef3f196974be06837373fb1f6f3c6a4e90279fb846", + "prepared_review_unit_hash": "dc54e8d59f280f0a43d590774fc4b3523f3a06f50bebe9ed567df94197419f71", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 36805, - "estimated_input_tokens": 9202, - "estimated_eval_tokens": 18660 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 36519, + "estimated_input_tokens": 9130, + "estimated_eval_tokens": 18516 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: the items cover unrelated tokenizer, import, model-loading, training, saving, and documentation/configuration bugs. None of the soft-edge pairs look like the same underlying bug or change, so all candidate duplicate links should be rejected.", - "confidence": 0.95, - "canonical_issue_reason": "No single canonical issue is safe here; the reports span unrelated problems across different models and code paths, so picking one would be arbitrary.", + "summary": "This cluster is highly heterogeneous. Most items are unrelated model-specific bugs or regressions that only share vague Transformer-version or import-related themes, so there are no strong duplicate pairs. The closest umbrella issue is the broad import-error report.", + "confidence": 0.9, + "canonical_issue_reason": "Issue 45278 is the broadest symptom-level report in the set, covering multiple import failures after an upgrade and making it the best umbrella for the small import-regression subgroup.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue stands out because there is no coherent duplicate cluster to anchor.", + "best_issue_reason": "45278 is the most triage-friendly general report here; it is broad enough to subsume several specific import errors, though it does not represent the many unrelated non-import bugs in the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:42907", - "right": "issue:44945", + "left": "issue:30990", + "right": "issue:33357", "accept": false, - "reason": "Different problems: saving dequantized Ministral/Devstral models vs incorrect output under pipeline parallelism." + "reason": "Different failures: loading hang vs Mac bus error on a CLIP model. Same vague area, not the same bug." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "AutoProcessor kwarg forwarding and chat-template crashes are different code paths and different symptoms." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 configuration guidance and a GlmMoeDsaIndexer torch.split issue are unrelated." }, { "left": "issue:43232", "right": "issue:45245", "accept": false, - "reason": "One is a generation sync/cache bug; the other is a categorical-cardinality runtime limit error." + "reason": "Generation sync/caching bug vs category-cardinality runtime error; no concrete overlap." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Saving dequantized models and incorrect pipeline-parallel outputs are different problems." }, { "left": "issue:44295", "right": "issue:44297", "accept": false, - "reason": "Position_ids buffer access error and tokenizer_class metadata mismatch are unrelated." + "reason": "Reading a registered buffer and tokenizer_class metadata mismatch are separate serialization issues." }, { "left": "issue:43232", "right": "issue:44743", "accept": false, - "reason": "Both involve generation state, but the concrete bugs differ: sync_gpus handling vs recurrent-state reset with cache." + "reason": "Both touch generation/cache behavior, but they describe different failures in different paths." }, { "left": "issue:43976", "right": "issue:44945", "accept": false, - "reason": "Python version/package compatibility issue vs incorrect output from pipeline parallelism." + "reason": "Python version compatibility and pipeline-parallel output corruption are unrelated." }, { "left": "issue:39401", "right": "issue:44488", "accept": false, - "reason": "Tokenizer offset_mapping regression and failure to load a specific model are different issues." + "reason": "Tokenizer offset mapping regression and model loading failure are not the same bug." }, { "left": "issue:42907", "right": "issue:43937", "accept": false, - "reason": "Saving dequantized models and invalid GenerationConfig are unrelated code paths." + "reason": "FP8 accuracy concerns and a GenerationConfig validation error do not match." }, { "left": "issue:43408", "right": "issue:43475", "accept": false, - "reason": "Configuration/model-type warning vs missing attribute on a SAM 3 video output object." + "reason": "Model-type warning and missing attribute in SAM 3 video are related only by subsystem, not by identical cause." }, { "left": "issue:44625", "right": "issue:44829", "accept": false, - "reason": "num_labels propagation and flash_attention_3 training degeneration are different bugs." + "reason": "num_labels propagation and flash_attention_3 training degeneration are different model/config issues." }, { "left": "issue:41628", "right": "issue:45278", "accept": false, - "reason": "A specific missing import and a broad set of post-upgrade import errors are not the same underlying bug." + "reason": "A single missing import and a broad upgrade-wide import-error report are not the same concrete defect." }, { - "left": "issue:29127", - "right": "issue:30990", + "left": "issue:44360", + "right": "issue:44485", "accept": false, - "reason": "LayoutLMv3 box-handling error message vs Sentence Transformers loading hang." + "reason": "DSA indexer activation behavior and GLM-5 RoPE implementation are unrelated." }, { - "left": "issue:44360", - "right": "issue:44485", + "left": "issue:29127", + "right": "issue:30990", "accept": false, - "reason": "DSA indexer missing ReLU and GLM-5 RoPE implementation are unrelated." + "reason": "Clarity of an error message vs a loading hang are not duplicate symptoms." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Example script multi-label output bug vs Gemma 4 mm_token_type_ids defaulting issue." + "reason": "Empty multi-label outputs in an example script and Gemma 4 mm_token_type_ids defaults are different issues." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Qwen3 offset_mapping bug and multimodal assistant_mask generation bug are distinct." + "reason": "Offset mapping bug and all-zero assistant masks in multimodal chat templating are separate tokenizer/template problems." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "Missing RMS norm epsilon and a wrong docstring for position_ids are unrelated." }, { "left": "issue:43122", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they involve different models and likely different fixes." + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "Qwen3-VL-Moe loading issue and tokenizer save metadata mismatch are unrelated." - }, - { - "left": "issue:44261", - "right": "issue:44373", - "accept": false, - "reason": "LayerNorm epsilon precision bug vs wrong docstring are not the same issue." + "reason": "Qwen3-VL-Moe loading in v5.0 and tokenizer save metadata mismatch are not the same bug." }, { "left": "issue:42371", "right": "issue:42831", "accept": false, - "reason": "TF32 behavior warning/update and FineGrainedFP8 accuracy regression are different topics." + "reason": "TF32 settings guidance and FineGrainedFP8 accuracy loss are different topics." }, { "left": "issue:44038", "right": "issue:44295", "accept": false, - "reason": "Qwen3-VL-Moe model issue and position_ids buffer error are unrelated." + "reason": "Model loading/regression and position_ids buffer access are unrelated." }, { "left": "issue:42907", "right": "issue:44964", "accept": false, - "reason": "Saving dequantized Ministral/Devstral models vs loading Phi-4 multimodal instruct." + "reason": "Quantized/dequantized save failure and Phi-4 multimodal loading failure do not share a concrete code-path." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Tensor-parallel training bug and multi-label classification example failure do not share a concrete code path." + "reason": "Tensor-parallel training bug and multi-label example script output bug are unrelated." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Dequantized model saving and gguf_file.from_pretrained breakage are different loading/saving paths." + "reason": "Saving dequantized models and gguf_file loading via AutoTokenizer/AutoModel are different serialization paths." }, { "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Qwen3-VL-Moe loading failure and FP8 save_pretrained on moe models are separate bugs." + "reason": "Qwen3-VL-Moe load bug and FP8 moe save_pretrained issue are different operations and likely different fixes." }, { "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "Missing embed_tokens.weight in UMT5Encoder and kwargs not passed in AutoProcessor.from_pretrained are unrelated." + "reason": "Missing embed_tokens.weight in UMT5Encoder loading and AutoProcessor kwarg forwarding are unrelated." }, { "left": "issue:43122", "right": "issue:44977", "accept": false, - "reason": "Both are tokenizer regressions, but there is no evidence they are the same tokenizer or same root cause." + "reason": "Different tokenizer regressions, on different models and with different symptoms." }, { "left": "issue:45216", "right": "issue:45478", "accept": false, - "reason": "Both concern Qwen3.5 regressions, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading failure; not enough to treat as the same bug." + "reason": "Save_pretrained checkpoint corruption and a from_pretrained error are not enough to treat as the same concrete defect." }, { "left": "issue:43278", "right": "issue:44945", "accept": false, - "reason": "Embedding dtype drift during evaluate and incorrect output under pipeline parallelism are different failures." + "reason": "Dtype drift between train/eval and incorrect pipeline-parallel outputs are separate bugs." }, { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Two different import errors for different symbols; same broad symptom, different root causes." - }, - { - "left": "issue:42907", - "right": "issue:44222", - "accept": false, - "reason": "Both mention saving models, but one is dequantized Ministral/Devstral and the other is FP8 MoE; too different to merge." - }, - { - "left": "issue:42757", - "right": "issue:44351", - "accept": false, - "reason": "Different missing imports from different packages/symbols." - }, - { - "left": "issue:42222", - "right": "issue:43441", - "accept": false, - "reason": "A broken vitpose module path and a FlashAttention failure for Ministral-3 are unrelated." - }, - { - "left": "issue:43784", - "right": "issue:45446", - "accept": false, - "reason": "Sentence-transformers import NameError and flex_attention version-check bug are unrelated." - }, - { - "left": "issue:44038", - "right": "issue:44295", - "accept": false, - "reason": "Qwen3-VL-Moe loading issue and position_ids buffer access error are different code paths." - }, - { - "left": "issue:44625", - "right": "issue:44829", - "accept": false, - "reason": "Missing num_labels propagation and flash_attention_3 training collapse are not the same defect." + "reason": "Two import errors, but for different symbols and likely different causes." }, { "left": "issue:42907", "right": "issue:44222", "accept": false, - "reason": "Both are save_pretrained-related, but the model families and failure modes differ." - }, - { - "left": "issue:45216", - "right": "issue:45478", - "accept": false, - "reason": "Related family, but the specific regression surfaces in save_pretrained vs from_pretrained and lacks evidence of a single shared root cause." - }, - { - "left": "issue:43278", - "right": "issue:44945", - "accept": false, - "reason": "Embedding dtype mismatch and pipeline-parallel output corruption are distinct issues." - }, - { - "left": "issue:43824", - "right": "issue:44351", - "accept": false, - "reason": "Import failures for different missing symbols are not duplicate bugs." + "reason": "Both involve saving models, but one is dequantized Ministrals/Devstrals and the other is FP8 MoE; not the same bug." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but they involve different names and likely different dependency-version breakages." + "reason": "Importing is_offline_mode from huggingface_hub and importing HybridCache from transformers are unrelated import failures." } ] }, @@ -11263,12 +11254,13 @@ "issue:33453", "issue:38617", "issue:39401", - "issue:41628", + "issue:42222", "issue:42371", "issue:42673", "issue:43064", "issue:43116", "issue:43421", + "issue:43441", "issue:43450", "issue:43493", "issue:43550", @@ -11307,6 +11299,8 @@ "issue:45563" ], "soft_pairs": [ + "issue:42222|issue:43441", + "issue:43784|issue:45446", "issue:42673|issue:45072", "issue:43064|issue:43673", "issue:22355|issue:29127", @@ -11330,221 +11324,226 @@ "issue:43827|issue:45446", "issue:42673|issue:43906", "issue:44291|issue:45071", - "issue:43450|issue:43986", "issue:45405|issue:45446", + "issue:43450|issue:43986", "issue:44829|issue:45216", - "issue:44484|issue:45310", "issue:39401|issue:43421", - "issue:38617|issue:41628" + "issue:44484|issue:45310" ], - "prepared_review_unit_hash": "b38649e4264a32859808178f7d75579815d7991551228dec5110f202b1283c0d", + "prepared_review_unit_hash": "0a3823f6403cfd829a1050a8c08a62ac37a73db301b8211b8d8abf28b269d4f2", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35530, - "estimated_input_tokens": 8883, - "estimated_eval_tokens": 18022 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36397, + "estimated_input_tokens": 9100, + "estimated_eval_tokens": 18456 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a grab-bag of unrelated Transformers issues, with one plausible duplicate family around v5 import regressions. The only soft edge I\u2019d accept is the specific `layer_type_validation` import failure against the broader \u201cmany import errors after update to 5.5.0\u201d report.", - "confidence": 0.72, - "canonical_issue_reason": "issue:45278 is the broadest and most representative issue in the only clear duplicate family here: post-upgrade v5 import failures. It can act as the umbrella for the specific import-name regression in issue:38617.", + "summary": "These items are mostly unrelated, model-specific regressions. The only near-duplicate I see is the configuration_utils import failure being a specific instance of the broader v5 import-regression report.", + "confidence": 0.67, + "canonical_issue_reason": "No single issue covers the whole set; if one must anchor the cluster, issue 45278 is the broadest umbrella for the import-regression subset.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 best fits the cluster because it captures a wider regression category rather than a single symptom, making it the most useful canonical issue if these are triaged together.", + "best_issue_reason": "Issue 45278 is the most representative issue here because it is broad enough to subsume the specific configuration_utils import failure in 38617.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Different models and failure modes: vitpose import breakage vs Ministral flash-attention error." + }, + { + "left": "issue:43784", + "right": "issue:45446", + "accept": false, + "reason": "Both are import-time problems, but they affect different modules and unrelated code paths." + }, { "left": "issue:42673", "right": "issue:45072", "accept": false, - "reason": "Different problems: VRAM leak in Qwen3ForCausalLM versus dtype mismatch in SwitchTransformers/TimmWrapperModel." + "reason": "VRAM leak in Qwen3 CausalLM threads is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:43064", "right": "issue:43673", "accept": false, - "reason": "Unrelated code paths: optimizer-state corruption under FSDP/PEFT versus GenerationMixin cache behavior in chunked prefill." + "reason": "Training/FSDP optimizer-state corruption and chunked-prefill cache absence are different bugs." }, { "left": "issue:22355", "right": "issue:29127", "accept": false, - "reason": "No shared bug: ONNX import failure versus a LayoutLMv3 error-message clarity issue." + "reason": "Module import failure for transformers.onnx is unrelated to a LayoutLMv3 error-message clarity issue." }, { "left": "issue:43550", "right": "issue:43784", "accept": false, - "reason": "Different failure modes and subsystems: torch.compile/SDPA on Bamba versus a sentence-transformers import NameError." + "reason": "torch.compile+SDPA failure for Bamba is unrelated to the sentence-transformers import NameError." }, { "left": "issue:38617", "right": "issue:45278", "accept": true, - "reason": "Both point to the same v5 import-regression family; 45278 is the umbrella report and 38617 is a specific missing-symbol import failure." + "reason": "38617 is a specific import error that fits the broader v5 import-regression report in 45278." }, { "left": "issue:43937", "right": "issue:44485", "accept": false, - "reason": "GLM-5 GenerationConfig validation error is distinct from a RoPE implementation discussion." + "reason": "GLM-5 generation-config validation and RoPE implementation are different parts of the model stack." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Different components and symptoms: classification example output bug versus SigLIP2 implementation discrepancy." + "reason": "Multi-label example-script output bug is unrelated to SigLIP2 implementation discrepancy." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 behavior guidance and GLM46V image-processor attribute error are unrelated." + "reason": "TF32 API guidance and missing image-patch helper are unrelated." }, { "left": "issue:44075", "right": "issue:44936", "accept": false, - "reason": "Optimizer arguments not being used is not the same as evaluate() failing after train()." + "reason": "SGD argument handling and trainer.evaluate() failure after train() are different code paths." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "Both mention Qwen3.5 MoE, but one is FP8 save_pretrained and the other is from_pretrained loading; not the same concrete bug." + "reason": "save_pretrained FP8/moe behavior is not the same bug as from_pretrained loading failure." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "Example-script multi-label output bug is unrelated to a pipeline() error-message regression." + "reason": "Classification example output is unrelated to pipeline translation-task messaging after v5 removals." }, { "left": "issue:44038", "right": "issue:45478", "accept": false, - "reason": "Related model family, but different concrete failures: Qwen3-VL-Moe bug versus Qwen3.5 MoE from_pretrained error." + "reason": "Different Qwen3-VL/Qwen3.5 MoE loading issues with no shared concrete failure described." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "Python version compatibility complaint is not the same as a PyTorch version check bug in flex_attention." + "reason": "Python version compatibility and PyTorch version gating are unrelated problems." }, { "left": "issue:43937", "right": "issue:45563", "accept": false, - "reason": "GenerationConfig invalidity and a stale warning for num_return_sequences are separate issues." + "reason": "GenerationConfig invalid and stale num_return_sequences warning are distinct generation issues." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer-related, but they affect different tokenizers and different output defects." + "reason": "Both involve tokenizers, but the titles point to different regressions and different models." }, { "left": "issue:43931", "right": "issue:44945", "accept": false, - "reason": "Model weight-shape mismatch on load is not the same as incorrect output under pipeline parallelism." + "reason": "Qwen3-VL weight-shape loading failure is unrelated to pipeline-parallel output corruption." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Tokenizer-loading regression and Qwen3 offset_mapping bug are too different to treat as duplicates." + "reason": "Tokenizer loading regression and Qwen3 offset_mapping bug are too different to treat as the same defect." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "A qwen3_omni_moe helper-function behavior issue is unrelated to a PyTorch version-check bug." + "reason": "Feature-extraction length helper behavior in qwen3_omni_moe is unrelated to flex_attention import version checks." }, { "left": "issue:43937", "right": "issue:44977", "accept": false, - "reason": "GLM-5 config validation and Qwen3.5 flash-attention generation problems are different failures." + "reason": "GenerationConfig validation and flash-attention generation failure are separate issues." }, { "left": "issue:45341", "right": "issue:45561", "accept": false, - "reason": "A testing_utils bug and an xdist race on captured_info.txt are separate defects, not the same root cause." + "reason": "Generic testing_utils bug and xdist race on captured_info.txt are not clearly the same underlying problem." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Docs still referencing pipeline() and a flex_attention version-check error are unrelated." + "reason": "Docs still referencing pipeline() and a flex_attention import check are unrelated." }, { "left": "issue:42673", "right": "issue:43906", "accept": false, - "reason": "VRAM leak under dataloader threads is unrelated to an isolated reproduction of another issue." + "reason": "The isolated reproduction target is unspecified here and does not match the VRAM-leak thread issue." }, { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "init_empty_weights argument handling and PretrainedConfig type checking are different compatibility regressions." - }, - { - "left": "issue:43450", - "right": "issue:43986", - "accept": false, - "reason": "Video processor batching shape bug is unrelated to AutoProcessor crashing without torchvision." + "reason": "init_empty_weights loading TypeError and PretrainedConfig type-checking breakage are different regressions." }, { "left": "issue:45405", "right": "issue:45446", "accept": false, - "reason": "PEFT version pinning and flex_attention version checking are unrelated dependency issues." + "reason": "Unreleased PEFT minimum-version bump is unrelated to flex_attention import gating." }, { - "left": "issue:44829", - "right": "issue:45216", + "left": "issue:43450", + "right": "issue:43986", "accept": false, - "reason": "Flash Attention 3 training degeneration is not the same as a save_pretrained checkpoint regression." + "reason": "Batched video-processor shape bug and missing torchvision crash are different problems." }, { - "left": "issue:44484", - "right": "issue:45310", + "left": "issue:44829", + "right": "issue:45216", "accept": false, - "reason": "A question about shard size defaults is not the same as a Qwen3.5 MoE loading error." + "reason": "flash_attention_3 training degeneration and save_pretrained checkpoint corruption are unrelated." }, { "left": "issue:39401", "right": "issue:43421", "accept": false, - "reason": "Tokenizer offset mapping and runtime post-processor updates are different tokenizer behaviors." + "reason": "Offset-mapping regression and runtime special-token post-processor update are plausibly related, but not clearly the same bug from the available evidence." }, { - "left": "issue:38617", - "right": "issue:41628", + "left": "issue:44484", + "right": "issue:45310", "accept": false, - "reason": "Both are import errors, but they concern different missing names and likely different regressions." + "reason": "max_shard_size discussion and Qwen3.5 from_pretrained failure are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary stays within the packet: it correctly treats the set as mostly unrelated issues and identifies 45278 as the broad umbrella for the import-regression family. The only accepted soft edge, 38617 vs 45278, is conservative and supported by the titles. No obvious evidence overstatement." + "feedback": "The cluster summary is conservative and grounded in the packet. The only clearly defensible near-duplicate is 38617 under the broader import-regression umbrella of 45278; the other soft-pair rejections are appropriately cautious based on the issue titles alone." }, "evaluator_used": true, "retried": false, @@ -11615,16 +11614,17 @@ "issue:45542" ], "soft_pairs": [ + "issue:38617|issue:41628", "issue:44117|issue:45081", - "issue:42673|issue:43582", "issue:44410|issue:45310", + "issue:42673|issue:43582", "issue:43408|issue:43742", "issue:42673|issue:43577", "issue:42222|issue:45542", "issue:42371|issue:44261", "issue:43937|issue:44496", - "issue:44483|issue:45464", "issue:38617|issue:44220", + "issue:44483|issue:45464", "issue:42175|issue:42222", "issue:42617|issue:45341", "issue:44933|issue:45278", @@ -11644,251 +11644,245 @@ "issue:45003|issue:45341", "issue:43931|issue:44410", "issue:44117|issue:44987", - "issue:43986|issue:45341", "issue:43097|issue:43452", + "issue:43986|issue:45341", "issue:43335|issue:45071", "issue:43931|issue:44038", "issue:43408|issue:44368", "issue:43874|issue:44263" ], - "prepared_review_unit_hash": "7df1b98165842a7eff8094d802bf22dc677647190889d212de32c38596b81195", + "prepared_review_unit_hash": "3f7497958e5b91fa3e3fce3e60c22a5c56032f61782cf02f864755c7aec54ebb", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 38020, - "estimated_input_tokens": 9505, - "estimated_eval_tokens": 19266 + "soft_pair_count": 36, + "serialized_chars": 38311, + "estimated_input_tokens": 9578, + "estimated_eval_tokens": 19412 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a mix of unrelated Transformers regressions and model-specific bugs. The only plausible umbrella theme is v5/v5.4/v5.5 import/API breakage, so the broadest issue is the best canonical issue candidate. None of the soft pairs look like the same concrete bug/change closely enough to merge.", - "confidence": 0.62, - "canonical_issue_reason": "issue:45278 is the broadest umbrella among the set: it describes many import errors after upgrading to v5.5.0, which fits the shared version-regression/import-failure theme better than the narrower reports.", + "summary": "No soft pairs look like clear same-bug duplicates under a strict triage standard; most are only loosely related by broad subsystem or version-regression themes. The strongest umbrella issue in the set is the broad import-regression report, but the evidence here is still too weak to merge any pair.", + "confidence": 0.34, + "canonical_issue_reason": "issue:45278 is the broadest representative of the visible cluster: it summarizes multiple import failures after a Transformers upgrade and is more likely to act as the umbrella report than the narrower, symbol-specific import issues.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best global issue candidate because it is the most general and reusable tracking item for the import-regression subset; the rest are mostly specific model/back-end bugs or unrelated API issues.", + "best_issue_reason": "issue:45278 is the best single issue to keep because it is broad, actionable, and likely covers the wider import-regression surface better than the more specific symptom reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44117", - "right": "issue:45081", + "left": "issue:38617", + "right": "issue:41628", "accept": false, - "reason": "Both are tokenizer-related, but one is a mapping-null bug and the other is a Mistral regex patch crash; different code paths and fixes." + "reason": "Both are import errors, but they concern different missing symbols and different entry points; no evidence they share the same underlying bug." }, { - "left": "issue:42673", - "right": "issue:43582", + "left": "issue:44117", + "right": "issue:45081", "accept": false, - "reason": "VRAM leak in Qwen3 CausalLM vs Apple Silicon TypeError in allocator warmup are unrelated failures." + "reason": "These hit different code paths: tokenizer mapping fallback logic versus a Mistral regex patch crash. Similar upgrade-era failure mode, but not the same bug." }, { "left": "issue:44410", "right": "issue:45310", "accept": false, - "reason": "Different Qwen model families and different symptoms: missing projections vs from_pretrained load error." + "reason": "Both involve Qwen-family model loading, but the reported failures are different and point to different model/config paths." }, { - "left": "issue:43408", - "right": "issue:43742", + "left": "issue:42673", + "right": "issue:43582", "accept": false, - "reason": "A model-type warning for sam3_video/sam3_tracker is unrelated to a MobileLLM key error." + "reason": "One is a VRAM leak in threaded dataloading; the other is an Apple Silicon TypeError in a warmup helper. No shared code-path evidence." }, { - "left": "issue:42673", - "right": "issue:43577", + "left": "issue:43408", + "right": "issue:43742", "accept": false, - "reason": "Memory leak in model use across threads is not the same as Blip2 dtype propagation staying float32." + "reason": "A SAM3 model-type warning and a MobileLLM loading KeyError are unrelated symptoms." }, { "left": "issue:42222", "right": "issue:45542", "accept": false, - "reason": "VitPose model breakage and missing TensorFlow backend are different subsystems and failure modes." + "reason": "VitPose breakage and a TensorFlow backend detection/install issue are separate problems." }, { "left": "issue:42371", "right": "issue:44261", "accept": false, - "reason": "TF32 settings warning is unrelated to the MLA rms_norm_eps precision/config issue." + "reason": "TF32 API guidance and an RMS norm epsilon precision bug are unrelated." }, { "left": "issue:43937", "right": "issue:44496", "accept": false, - "reason": "GLM generation-config validation and an unrecognized model config are separate loading/validation problems." + "reason": "Different model-loading failures with different causes: invalid GenerationConfig versus unrecognized model config." }, { - "left": "issue:44483", - "right": "issue:45464", + "left": "issue:38617", + "right": "issue:44220", "accept": false, - "reason": "Both hit chat/completions, but one is a generic request acceptance issue and the other is a streaming inference failure on a specific model." + "reason": "An import error and an audio feature extraction issue do not look like the same underlying bug." }, { - "left": "issue:38617", - "right": "issue:44220", + "left": "issue:44483", + "right": "issue:45464", "accept": false, - "reason": "An import error for layer_type_validation is unrelated to an audio feature extraction bug." + "reason": "Both concern chat/completions behavior, but one is a general API rejection and the other is a streaming inference failure; not enough to treat as the same bug." }, { "left": "issue:42175", "right": "issue:42222", "accept": false, - "reason": "TensorFlow backend packaging and VitPose model breakage are not the same issue." + "reason": "TensorFlow installation/backend packaging and VitPose model breakage are not the same issue." }, { "left": "issue:42617", "right": "issue:45341", "accept": false, - "reason": "A script execution problem and a testing_utils bug are unrelated." + "reason": "A runtime script failure and a testing utility bug are unrelated." }, { "left": "issue:44933", "right": "issue:45278", "accept": false, - "reason": "The latter is a broad import-regression umbrella, but this specific missing image_utils import is not clearly the same concrete bug." + "reason": "Both mention import problems, but one is a specific missing import from image_utils while the other is a broad post-upgrade import regression; insufficient evidence of one shared root cause." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "save_pretrained shard-size behavior is unrelated to the Qwen3.5 MoE loading error." + "reason": "These are unrelated: max_shard_size behavior versus a Qwen3.5 MoE from_pretrained failure." }, { "left": "issue:43066", "right": "issue:43122", "accept": false, - "reason": "Both are tokenizer/v5-related, but one is a decoder-type mismatch and the other is a tokenization behavior change; not clearly the same fix." + "reason": "Both involve tokenization changes in v5, but the reported symptoms are different enough that the shared root cause is not established." }, { "left": "issue:43673", "right": "issue:45341", "accept": false, - "reason": "Generation cache behavior during chunked prefill is unrelated to a testing utility bug." + "reason": "Generation cache handling and a testing-utils bug do not match." }, { "left": "issue:43097", "right": "issue:44493", "accept": false, - "reason": "Removing tie_embeddings_and_encoder_decoder and unexpected position-id keys are different regressions." + "reason": "Removed tie_embeddings support and unexpected position-id keys are different config/modeling regressions." }, { "left": "issue:44410", "right": "issue:45478", "accept": false, - "reason": "Different Qwen model-loading failures with different root causes." + "reason": "Different Qwen-family failures with different model variants and symptoms; not the same concrete bug." }, { "left": "issue:42222", "right": "issue:45459", "accept": false, - "reason": "VitPose model breakage and tokenizer error handling with missing protobuf are unrelated." + "reason": "VitPose model breakage and tokenizer error masking from protobuf handling are unrelated." }, { "left": "issue:43874", "right": "issue:44261", "accept": false, - "reason": "Missing get_number_of_image_patches in an image processor is unrelated to the MLA rms_norm_eps issue." + "reason": "Different multimodal/model config issues with no clear shared code-path." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "FlashAttention failures, but on different models and code paths; not the same concrete bug." + "reason": "Both mention FlashAttention, but one is a Ministral loading failure and the other is degenerate training; not the same fix." }, { "left": "issue:43986", "right": "issue:44484", "accept": false, - "reason": "AutoProcessor/video-model import behavior is unrelated to max_shard_size in save_pretrained." + "reason": "Video AutoProcessor import/torchvision behavior is unrelated to max shard sizing." }, { "left": "issue:43012", "right": "issue:43408", "accept": false, - "reason": "A PyTorch bfloat16 warning and a sam3 model-type warning do not share a root cause." + "reason": "A precision warning during compile and a SAM3 model-type warning are unrelated." }, { "left": "issue:44263", "right": "issue:44485", "accept": false, - "reason": "A torch.split return-value problem and a GLM-5 RoPE implementation issue are distinct." + "reason": "A torch.split return-value issue and a GLM RoPE implementation discussion are not the same bug." }, { "left": "issue:41628", "right": "issue:45071", "accept": false, - "reason": "Cannot import AutoImageProcessor and PreTrainedConfig type-checking breakage are different version regressions." + "reason": "An import error and a PretrainedConfig type-checking regression are distinct issues." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "PretrainedConfig type-checking and TensorFlow backend detection are unrelated." + "reason": "Type-checking and TensorFlow backend-install behavior are unrelated." }, { "left": "issue:45003", "right": "issue:45341", "accept": false, - "reason": "Unsafe sys.modules access in modeling_utils is unrelated to the testing_utils bug." + "reason": "Unsafe sys.modules access and a testing utility bug are not the same code-path." }, { "left": "issue:43931", "right": "issue:44410", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch and qwen3next missing projections are different model-loading failures." + "reason": "Both involve model loading, but one is a weight-shape mismatch and the other is missing projections in a different model family." }, { "left": "issue:44117", "right": "issue:44987", "accept": false, - "reason": "Tokenizer mapping None and a transformers>=5.1.0 load failure are not the same concrete bug." + "reason": "Both are from_pretrained-related failures, but the tokenizer mapping bug and the physical-intelligence/fast loading failure do not clearly share a root cause." }, { - "left": "issue:43986", - "right": "issue:45341", + "left": "issue:43097", + "right": "issue:43452", "accept": false, - "reason": "Video AutoProcessor crash without torchvision is unrelated to a testing utility bug." + "reason": "Tie-embeddings removal and gguf_file loading breakage are different regression classes." }, { - "left": "issue:43097", - "right": "issue:43452", + "left": "issue:43986", + "right": "issue:45341", "accept": false, - "reason": "Embedding-tie removal and gguf_file breaking from_pretrained are unrelated regressions." + "reason": "Video processor dependency handling and a testing utility bug are unrelated." }, { "left": "issue:43335", "right": "issue:45071", "accept": false, - "reason": "SwitchTransformers sparse-layer creation and PreTrainedConfig type checking are different issues." + "reason": "SwitchTransformers sparse-layer construction and config type-checking are separate problems." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Qwen3-VL weight mismatch and Qwen3-VL-Moe v5 bug are model-specific and distinct." - }, - { - "left": "issue:43408", - "right": "issue:44368", - "accept": false, - "reason": "Model-type warning for sam3 is unrelated to tie_word_embeddings warning during Qwen3.5 LoRA fine-tuning." + "reason": "Same broad Qwen3-VL area, but different symptoms and likely different causes." }, { "left": "issue:43874", "right": "issue:44263", "accept": false, - "reason": "Both touch GLM-related code, but one is an image-processor API omission and the other is a tensor-splitting issue; not the same bug." + "reason": "Different GLM-related failures with no evidence of the same underlying defect." } ] }, @@ -11929,6 +11923,7 @@ "issue:43531", "issue:43550", "issue:43618", + "issue:43824", "issue:43825", "issue:43873", "issue:43901", @@ -11955,7 +11950,6 @@ "issue:45071", "issue:45103", "issue:45245", - "issue:45290", "issue:45341", "issue:45372", "issue:45375", @@ -11971,17 +11965,17 @@ "issue:44261|issue:44485", "issue:44410|issue:45375", "issue:44062|issue:44117", - "issue:44038|issue:44410", "issue:44704|issue:45103", + "issue:44038|issue:44410", "issue:38617|issue:44351", "issue:43986|issue:44117", "issue:43502|issue:43992", "issue:43097|issue:43873", - "issue:43986|issue:45290", "issue:43992|issue:44484", + "issue:43824|issue:43986", "issue:43066|issue:43906", - "issue:43825|issue:44655", "issue:43352|issue:44380", + "issue:43825|issue:44655", "issue:38617|issue:43097", "issue:43097|issue:43901", "issue:43012|issue:43550", @@ -11997,52 +11991,52 @@ "issue:44483|issue:44514", "issue:43618|issue:44380" ], - "prepared_review_unit_hash": "7d948642efe4f0fe1e85a2de1fa96f9a8ae3cb9fdf7fe65f3e828faf95d09cce", + "prepared_review_unit_hash": "0edf23d0cdcaf93b23a3e0bd2e5f665db51e3005c5b1359a0887f6b371aa6907", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37834, - "estimated_input_tokens": 9459, - "estimated_eval_tokens": 19174 + "serialized_chars": 37827, + "estimated_input_tokens": 9457, + "estimated_eval_tokens": 19170 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is broadly heterogeneous: the soft pairs are mostly superficially similar (imports, tokenizer/model bugs, training issues, warnings) but do not describe the same underlying defect. I would not merge any of the proposed soft edges, and there is no clear single canonical issue for the whole set.", - "confidence": 0.94, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: the soft pairs mostly connect unrelated regressions, import errors, model-specific bugs, and docs/API mismatches. I reject all soft edges and do not see a true duplicate cluster here.", + "confidence": 0.95, + "canonical_issue_reason": "No single issue is a clear duplicate anchor for the whole set. If forced to pick a representative, issue 43097 is the clearest user-facing v5 regression with broad downstream impact and a concrete symptom.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 43097 is the strongest standalone representative: it is specific, reproducible, and affects a core v5 API change. It is the best candidate only as an anchor, not because the surrounding issues are duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44829", "right": "issue:45245", "accept": false, - "reason": "Different problems: flash-attention-3 training degeneration vs a category-cardinality runtime limit." + "reason": "Different failures: flash-attention training degeneration vs a categorical cardinality limit error." }, { "left": "issue:43986", "right": "issue:45372", "accept": false, - "reason": "Both are loading/import failures, but one is about missing torchvision for video models and the other is a mistral_common import breakage for Gemma 4 processing." + "reason": "Both are loading/import problems, but they involve different missing dependencies and different processor paths." }, { "left": "issue:43976", "right": "issue:45071", "accept": false, - "reason": "Different compatibility regressions: Python 3.9+ support vs PretrainedConfig type checking in v5.4.0." + "reason": "One is a Python version compatibility complaint; the other is a PretrainedConfig type-checking regression." }, { "left": "issue:43531", @@ -12054,175 +12048,175 @@ "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "Different training-stack issues: FSDP2/PEFT optimizer state corruption vs Liger kernel not applied with model_init." + "reason": "Trainer/FSDP optimizer-state corruption is a different code path from Liger Kernel not being applied with model_init." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they target different symbols from different modules and have no shared code path." + "reason": "Different import errors from different symbols/modules; same symptom class, not the same bug." }, { "left": "issue:44261", "right": "issue:44485", "accept": false, - "reason": "MLA layernorm precision/config bug is unrelated to GLM-5 RoPE implementation." + "reason": "Different model internals: MLA RMS norm precision issue vs GLM-5 RoPE implementation." }, { "left": "issue:44410", "right": "issue:45375", "accept": false, - "reason": "Different Qwen-family model config issues affecting different fields and components." + "reason": "Both are Qwen3-family issues, but one is missing projections and the other is a missing config field." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "Tokenizer construction error and tokenizer-mapping None handling are distinct failures." + "reason": "Tokenizer construction error vs tokenizer mapping fallback bug; not the same concrete failure." }, { - "left": "issue:44038", - "right": "issue:44410", + "left": "issue:44704", + "right": "issue:45103", "accept": false, - "reason": "Both mention Qwen, but they concern different model variants and different missing model pieces." + "reason": "AutoProcessor kwarg forwarding and auto_docstring annotation handling are unrelated code paths." }, { - "left": "issue:44704", - "right": "issue:45103", + "left": "issue:44038", + "right": "issue:44410", "accept": false, - "reason": "AutoProcessor kwargs forwarding is unrelated to the auto_docstring future-annotations crash." + "reason": "Distinct Qwen model bugs with different symptoms and fixes." }, { "left": "issue:38617", "right": "issue:44351", "accept": false, - "reason": "Both are import-related, but they involve different missing APIs and different subsystems." + "reason": "Both are import errors, but they refer to different missing public symbols." }, { "left": "issue:43986", "right": "issue:44117", "accept": false, - "reason": "Video-model loading without torchvision is unrelated to tokenizer mapping returning None." + "reason": "Video-model processor loading without torchvision is unrelated to tokenizer mapping returning None." }, { "left": "issue:43502", "right": "issue:43992", "accept": false, - "reason": "local_files_only still making network calls is unrelated to a missing embed_tokens.weight on UMT5Encoder load." + "reason": "Local-files-only network leakage is unrelated to a missing weight during UMT5Encoder loading." }, { "left": "issue:43097", "right": "issue:43873", "accept": false, - "reason": "A removed config field warning is unrelated to offloading/quantization behavior." + "reason": "Tie-embeddings deprecation/regression is unrelated to quantization offloading behavior." }, { - "left": "issue:43986", - "right": "issue:45290", + "left": "issue:43992", + "right": "issue:44484", "accept": false, - "reason": "Different processor failures: missing torchvision vs batched chat template handling with tool-call messages." + "reason": "Missing embed_tokens weight on load and save_pretrained shard-size behavior are different operations." }, { - "left": "issue:43992", - "right": "issue:44484", + "left": "issue:43824", + "right": "issue:43986", "accept": false, - "reason": "Model weight-loading failure and save_pretrained sharding default are unrelated." + "reason": "Importing a missing model class is not the same issue as crashing on AutoProcessor without torchvision." }, { "left": "issue:43066", "right": "issue:43906", "accept": false, - "reason": "Wrong tokenizer decoder type is not the same bug as an isolated reproduction of another issue." + "reason": "Tokenizer decoder-type regression is too different from an isolated reproduction of another issue." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Unsupported FlashAttention 2 warning for Nemotron is different from GPT2 attention scaling being ignored." }, { "left": "issue:43825", "right": "issue:44655", "accept": false, - "reason": "Pipeline task-support messaging and pipeline save_pretrained failure are separate issues." + "reason": "Pipeline translation-message wording and saving Pipeline objects are separate behaviors." }, { "left": "issue:38617", "right": "issue:43097", "accept": false, - "reason": "ImportError for a utility symbol and a removed config option are unrelated." + "reason": "ImportError for a missing helper and a removed config field are unrelated." }, { "left": "issue:43097", "right": "issue:43901", "accept": false, - "reason": "Tie-embeddings removal and outdated docs about return_all_scores are different changes." + "reason": "Removed tie_embeddings_and_encoder_decoder and return_all_scores doc drift do not share the same bug." }, { "left": "issue:43012", "right": "issue:43550", "accept": false, - "reason": "Different failures: a bfloat16 warning during compile vs a Bamba SDPA torch.compile bug." + "reason": "Compile-time warning on bfloat16 differs from a functional torch.compile+SDPA failure." }, { "left": "issue:43012", "right": "issue:44368", "accept": false, - "reason": "They are both warnings, but about different config fields and unrelated code paths." + "reason": "Precision warning during compile is unrelated to the tie_word_embeddings warning in LORA fine-tuning." }, { "left": "issue:42886", "right": "issue:43381", "accept": false, - "reason": "Offline cache loading and gradient-checkpointing eval-mode behavior are unrelated." - }, - { - "left": "issue:43208", - "right": "issue:43550", - "accept": false, - "reason": "xLSTM training bugs are unrelated to Bamba SDPA/torch.compile issues." + "reason": "Offline cache loading and eval-mode gradient checkpointing are unrelated." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Saving dequantized models and saving Pipeline objects both involve save_pretrained, but they are different object types and failures." + "reason": "Saving dequantized models is a different problem from saving Pipeline objects." }, { "left": "issue:43525", "right": "issue:44117", "accept": false, - "reason": "Missing pad_token_id on Llama4Config is unrelated to tokenizer mapping assumptions." + "reason": "Llama4Config missing pad_token_id is unrelated to tokenizer mapping assumptions." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "GPT2 attention scaling under SDPA/FlashAttention is unrelated to degenerate zero loss in single-label classification." + "reason": "Attention backend scaling bug and degenerate classification loss with num_labels=1 are unrelated." }, { "left": "issue:43352", "right": "issue:43976", "accept": false, - "reason": "Flash Attention support for Nemotron is unrelated to the Python-version compatibility issue." + "reason": "Model backend support and Python version compatibility are not the same underlying issue." }, { "left": "issue:43064", "right": "issue:44380", "accept": false, - "reason": "FSDP2 optimizer-state corruption and GPT2 attention scaling are different bugs." + "reason": "Optimizer-state corruption under FSDP2/PEFT is unrelated to GPT2 attention scaling." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "xLSTM training bugs and BeitImageProcessorFast reduce_labels behavior are unrelated." + "reason": "xLSTM training blockers and BeitImageProcessorFast label reduction are unrelated." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "Chat/completions API rejection and Qwen2_5_VLProcessor batching are different layers and different failures." + "reason": "Server endpoint request rejection and processor chat-template batching crash are different layers and failures." }, { "left": "issue:43618", "right": "issue:44380", "accept": false, - "reason": "Missing CLIPOutput attentions assignment is not the same as GPT2 attention scaling being ignored." + "reason": "CLIPOutput attention assignment regression is unrelated to GPT2 attention scaling under backend kernels." } ] }, @@ -12275,7 +12269,6 @@ "issue:44038", "issue:44077", "issue:44188", - "issue:44220", "issue:44230", "issue:44242", "issue:44246", @@ -12293,15 +12286,16 @@ "issue:45310", "issue:45375", "issue:45381", - "issue:45440" + "issue:45440", + "issue:45478" ], "soft_pairs": [ "issue:43618|issue:43761", "issue:36683|issue:43582", - "issue:43012|issue:44683", "issue:44483|issue:45290", - "issue:42757|issue:43824", + "issue:43012|issue:44683", "issue:44038|issue:44230", + "issue:42757|issue:43824", "issue:44483|issue:45381", "issue:43299|issue:43867", "issue:38617|issue:43502", @@ -12310,17 +12304,17 @@ "issue:42994|issue:44655", "issue:36683|issue:43475", "issue:43867|issue:45216", - "issue:43352|issue:43937", "issue:43066|issue:45381", - "issue:44220|issue:44655", + "issue:43352|issue:43937", + "issue:44655|issue:45478", "issue:44077|issue:44380", "issue:36683|issue:44077", - "issue:43064|issue:44492", "issue:44263|issue:45440", + "issue:43064|issue:44492", "issue:43976|issue:44938", "issue:36683|issue:43525", - "issue:44188|issue:44380", "issue:36010|issue:42757", + "issue:44188|issue:44380", "issue:43618|issue:45216", "issue:36683|issue:44242", "issue:43992|issue:45310", @@ -12331,238 +12325,244 @@ "issue:43867|issue:45375", "issue:36683|issue:43064" ], - "prepared_review_unit_hash": "23dafea54c6bbd124c865298d715dcfe360d3536894086cf15600ad4ec0bf84a", + "prepared_review_unit_hash": "ceebbf64f0b69fdd80c7df3b57734f75e17a7b50f8308f14e9658971b7b7db0a", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37407, - "estimated_input_tokens": 9352, - "estimated_eval_tokens": 18960 + "serialized_chars": 37419, + "estimated_input_tokens": 9355, + "estimated_eval_tokens": 18966 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: many look like unrelated v5 regressions, import/export issues, model-specific loading bugs, or documentation/performance complaints. I only see one plausible duplicate/soft-edge match around slow `import transformers` behavior.", - "confidence": 0.66, - "canonical_issue_reason": "Issue 45278 is the broadest umbrella in the set for the import-regression theme (\u201cmany import errors after update to 5.5.0\u201d) and is the most useful representative issue among these mostly mixed reports.", + "summary": "Mostly heterogeneous issue cluster with one clear duplicate family around import-time slowdown. One soft pair looks like the same import-performance bug; the rest are separate model/config/version-specific regressions.", + "confidence": 0.71, + "canonical_issue_reason": "issue:45216 is the clearest concrete regression in the Qwen3.5 MoE save/load family and reads like the best anchor for nearby reports about the same checkpoint/from_pretrained failure mode.", "canonical_pr_reason": null, - "best_issue_reason": "45278 is the most central and actionable issue for triage: it is open, broad, and covers a recurring upgrade/import failure pattern that overlaps several other reports in the cluster.", + "best_issue_reason": "issue:45216 is the most specific, actionable report among the issues here; it names the affected model, API, and version boundary.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43618", "right": "issue:43761", "accept": false, - "reason": "Both are CLIP regressions, but they hit different outputs (`attentions` vs `hidden_states`) and likely different code paths." + "reason": "Both are CLIP/v5 regressions, but one is about attentions and the other hidden_states; different concrete failures." }, { "left": "issue:36683", "right": "issue:43582", "accept": false, - "reason": "Gemma3 config attribute error vs Apple Silicon allocator TypeError; unrelated subsystems." - }, - { - "left": "issue:43012", - "right": "issue:44683", - "accept": false, - "reason": "A PyTorch precision warning is not the same bug as a compiled flex_attention failure." + "reason": "Unrelated failures: missing Gemma3Config vocab_size vs Apple Silicon caching allocator TypeError." }, { "left": "issue:44483", "right": "issue:45290", "accept": false, - "reason": "One is API request acceptance at `/v1/chat/completions`; the other is chat-template tokenization crashing on tool-call messages." + "reason": "Different paths: chat/completions request validation vs apply_chat_template crashing on tool-call messages." }, { - "left": "issue:42757", - "right": "issue:43824", + "left": "issue:43012", + "right": "issue:44683", "accept": false, - "reason": "Both are import errors, but for different missing symbols from different packages/modules." + "reason": "Both touch compilation/warnings, but one is a bfloat16 precision warning and the other is a torch>=2.9 flex_attention failure." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "Model loading regression vs fp8 inference support request; not the same concrete bug." + "reason": "Same model family, but one is a loading bug and the other is an fp8 inference support request; not the same bug." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and different breakpoints." }, { "left": "issue:44483", "right": "issue:45381", "accept": false, - "reason": "Chat completion request rejection and Qwen2.5-VL video position-id mismatch are different failures." + "reason": "Different bugs: API request rejection vs Qwen2.5-VL video position ids being wrong." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Different load failures: Qwen3-VL-Moe model loading vs sorted-state-dict loading error." + "reason": "Both are model-loading errors, but the affected models and failure modes differ." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "Missing import from `configuration_utils` vs unexpected network requests under `local_files_only=True`; unrelated." + "reason": "ImportError from configuration_utils is unrelated to local_files_only still allowing network requests." }, { "left": "issue:43756", "right": "issue:44485", "accept": false, - "reason": "Different model families and different RoPE-related problems; not the same bug." + "reason": "Both mention RoPE, but they target different models and different implementation concerns." }, { "left": "issue:43901", "right": "issue:44655", "accept": false, - "reason": "Docs mentioning `return_all_scores` vs inability to save Pipeline objects; different issues." + "reason": "Docs mismatch for return_all_scores is unrelated to saving Pipeline objects." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Both involve saving, but one is quantized model saving and the other is Pipeline serialization; different code paths." + "reason": "Quantized model saving and Pipeline.save_pretrained are different save-path bugs." }, { "left": "issue:36683", "right": "issue:43475", "accept": false, - "reason": "Missing `vocab_size` on Gemma3Config vs missing `fpn_position_embeddings` on SAM 3 output; unrelated attribute bugs." + "reason": "Different attribute-missing regressions in different model configs/outputs." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "A load-time state_dict ordering error and a save_pretrained regression for Qwen3.5 are different problems." + "reason": "Both are Qwen-related load/save problems, but one is a shape-mismatch load error and the other is a checkpoint correctness regression." }, { - "left": "issue:43352", - "right": "issue:43937", + "left": "issue:43066", + "right": "issue:45381", "accept": false, - "reason": "Flash Attention 2 support gating vs invalid GenerationConfig validation; different failure points." + "reason": "Tokenizer decoder type issue is unrelated to Qwen2.5-VL video position ids." }, { - "left": "issue:43066", - "right": "issue:45381", + "left": "issue:43352", + "right": "issue:43937", "accept": false, - "reason": "Tokenizer decoder type issue vs Qwen2.5-VL video vision-position-id bug; not the same underlying defect." + "reason": "Different model/runtime issues: Flash Attention support vs GenerationConfig validation." }, { - "left": "issue:44220", - "right": "issue:44655", + "left": "issue:44655", + "right": "issue:45478", "accept": false, - "reason": "Audio feature extraction bug vs Pipeline save_pretrained failure; unrelated." + "reason": "Pipeline save_pretrained and Qwen3.5 MoE from_pretrained are separate failure paths." }, { "left": "issue:44077", "right": "issue:44380", "accept": false, - "reason": "PatchTSMixer post-init validation and GPT2 attention scaling under SDPA/FlashAttention are different bugs." + "reason": "Different code paths: post_init validation vs attention scaling under SDPA/FlashAttention." }, { "left": "issue:36683", "right": "issue:44077", "accept": false, - "reason": "Gemma3 config missing field vs PatchTSMixer post_init policy; unrelated." + "reason": "Gemma3Config missing vocab_size is unrelated to patchtsmixer post_init rules." }, { - "left": "issue:43064", - "right": "issue:44492", + "left": "issue:44263", + "right": "issue:45440", "accept": false, - "reason": "Distributed optimizer-state corruption vs a typo in cache strategies; unrelated." + "reason": "Different model implementations and different correctness issues." }, { - "left": "issue:44263", - "right": "issue:45440", + "left": "issue:43064", + "right": "issue:44492", "accept": false, - "reason": "Different MoE implementations and different problems (`torch.split` indexing vs DeepSeekV3 divergence)." + "reason": "Distributed optimizer-state bug is unrelated to a cache-strategy typo." }, { "left": "issue:43976", "right": "issue:44938", "accept": false, - "reason": "Both are Python-version compatibility problems, but for different versions and different breakages." + "reason": "Both involve Python compatibility, but they are distinct version/load failures." }, { "left": "issue:36683", "right": "issue:43525", "accept": false, - "reason": "Both are missing config attributes, but for different models and different fields." + "reason": "Different configs missing different attributes; not the same bug." }, { - "left": "issue:44188", - "right": "issue:44380", + "left": "issue:36010", + "right": "issue:42757", "accept": false, - "reason": "Compile-time attention-kernel divergence is not the same issue as GPT2 attention scaling being ignored." + "reason": "Separate import errors for different symbols and causes." }, { - "left": "issue:36010", - "right": "issue:42757", + "left": "issue:44188", + "right": "issue:44380", "accept": false, - "reason": "Both are import errors, but they concern different missing names and different causes." + "reason": "Attention-kernel divergence under compile is a different bug from GPT2 attention scaling being ignored." }, { "left": "issue:43618", "right": "issue:45216", "accept": false, - "reason": "CLIP attention output regression and Qwen3.5 save_pretrained regression are unrelated." + "reason": "CLIP output regression is unrelated to Qwen3.5 save/load behavior." }, { "left": "issue:36683", "right": "issue:44242", "accept": false, - "reason": "Gemma3 config attribute error vs MoE load-balancing loss logic; unrelated." + "reason": "Missing config attribute vs MoE load-balancing loss omission are unrelated." }, { "left": "issue:43992", "right": "issue:45310", "accept": false, - "reason": "UMT5 weight-loading omission vs Qwen3.5 MoE from_pretrained error; different loading bugs." + "reason": "Different from_pretrained problems on different model families." }, { "left": "issue:43010", "right": "issue:45278", "accept": false, - "reason": "Cache update should be `no_grad` is unrelated to broad import errors after upgrade." + "reason": "Cache update no_grad decoration is unrelated to broad import errors after upgrade." }, { "left": "issue:36683", "right": "issue:43352", "accept": false, - "reason": "Missing config field vs unsupported Flash Attention 2 for Nemotron; different issues." + "reason": "Config attribute error and Flash Attention support error are different failures." + }, + { + "left": "issue:43352", + "right": "issue:43931", + "accept": false, + "reason": "Nemotron Flash Attention support is unrelated to Qwen3-VL weight-shape mismatch." }, { "left": "issue:41669", "right": "issue:44246", "accept": true, - "reason": "Both describe slow `import transformers`; 41669 points to the root cause (`import *` in model imports) and 44246 reports the same symptom." + "reason": "Same underlying import-time slowdown: one report identifies the model import-star cause, the other reports the symptom that importing transformers is slow." }, { "left": "issue:43867", "right": "issue:45375", "accept": false, - "reason": "State-dict loading error vs missing `deepstack_visual_indexes` in Qwen3.5 vision config; different bugs." + "reason": "Weight-shape mismatch on load is unrelated to a missing deepstack_visual_indexes field being dropped." }, { "left": "issue:36683", "right": "issue:43064", "accept": false, - "reason": "A config attribute error and a distributed optimizer-state bug are unrelated." + "reason": "Different subsystems and different failure modes." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The cluster is indeed heterogeneous, and the only soft-edge acceptance (41669 \u2194 44246) is plausible from the titles alone without overcommitting beyond the evidence. The other pairwise rejections also read as cautious and well-supported." + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The only accepted soft pair (issue:41669 / issue:44246) is supported by the shared import-performance symptom, and the other pairwise rejections do not overclaim identity. The Qwen3.5 MoE regression anchor at issue:45216 is a reasonable choice for the nearby save/pretrained reports." }, "evaluator_used": true, "retried": false, @@ -12660,8 +12660,8 @@ "issue:42548|issue:44351", "issue:45125|issue:45375", "issue:44355|issue:44749", - "issue:30990|issue:31795", "issue:43502|issue:44485", + "issue:30990|issue:31795", "issue:44075|issue:44871", "issue:44315|issue:44360", "issue:43010|issue:44368", @@ -12669,238 +12669,244 @@ "issue:43502|issue:44222", "issue:43299|issue:44556" ], - "prepared_review_unit_hash": "e18562543ce73575023181352eca3180288cc120e48465d0afdcbbb376d621eb", + "prepared_review_unit_hash": "41410ec29671a3804d8b3b57045d3d98c5383941cac953ba796e4c44da143b35", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 37636, - "estimated_input_tokens": 9409, - "estimated_eval_tokens": 19074 + "serialized_chars": 37642, + "estimated_input_tokens": 9411, + "estimated_eval_tokens": 19078 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous cluster of unrelated issues; the soft pairs are mostly same-subsystem at best, not true duplicates. I did not find any pair that clearly shares the same concrete bug/change, so all soft edges are rejected. The most representative issue is the broad checkpoint reload regression after upgrading to v5.x.", - "confidence": 0.48, - "canonical_issue_reason": "issue:44556 is the broadest, most central loading-regression report in the set: checkpoints trained on v4.57 fail to reload in v5.2/v5.3. It generalizes better than model-specific, tokenizer-specific, or test/doc-only reports.", + "summary": "This cluster is mostly a collection of superficially similar Transformers issues (loading, version regressions, tensor-parallel, docs, tests) but not duplicates of the same underlying bug. I would not merge any of the soft-edge pairs.", + "confidence": 0.95, + "canonical_issue_reason": "issue:43723 is the most broadly framed, concrete loading-regression report in the set (AutoTokenizer.from_pretrained in v5), making it the best representative issue among these superficially related reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44556 best captures the cluster\u2019s main recurring theme of v5 loading regressions without being overly narrow to one model or API path.", + "best_issue_reason": "issue:43723 is a clear, actionable model-loading regression with a specific API surface and version context; it is the strongest general representative of the loading-related subset in this cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "Both are test-flakiness reports, but they involve different failures and files; no shared underlying bug is evident." + "reason": "Different test bugs: one is about missing set_seed in SAM HQ integration tests, the other is a pytest-xdist race on captured_info.txt." }, { "left": "issue:43992", "right": "issue:45478", "accept": false, - "reason": "Different models and different loading failures; too vague to be the same code-path bug." + "reason": "Both are loading-related, but one is a missing embed_tokens.weight in UMT5Encoder and the other is a Qwen3.5 MoE from_pretrained error; different model/code paths." }, { "left": "issue:44938", "right": "issue:45071", "accept": false, - "reason": "One is a Python 3.14 load failure, the other is a PretrainedConfig type-checking regression; not the same bug." + "reason": "Unrelated issues: Python 3.14 load failure vs PretrainedConfig type-checking breakage in v5.4.0." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "Both concern loading/saving keys, but they affect different model families and different directions of the flow." + "reason": "Different failure modes: sorted state_dict loading error vs incorrect visual encoder keys saved by Qwen3.5 save_pretrained." }, { "left": "issue:43010", "right": "issue:43976", "accept": false, - "reason": "Completely unrelated: no-grad decoration on cache updates vs Python-version compatibility." + "reason": "No shared bug: one is about @torch.no_grad on cache/layer updates, the other is a Python-version compatibility problem." }, { "left": "issue:44938", "right": "issue:45381", "accept": false, - "reason": "Generic load failure vs a Qwen2.5-VL video position_ids bug; different problems." + "reason": "Different model/version bugs: Python 3.14 import/load failure vs wrong vision_position_ids for qwen2.5-vl video input." }, { "left": "issue:44360", "right": "issue:45440", "accept": false, - "reason": "Both mention implementation divergence, but for different models and different behaviors; not a duplicate." + "reason": "Completely different areas: DSA indexer ReLU discussion vs DeepseekV3MoE implementation divergence." }, { "left": "issue:43618", "right": "issue:44077", "accept": false, - "reason": "Unrelated model-output field issue vs an optional post_init constraint." + "reason": "Different components: CLIPOutput attentions regression vs patchtsmixer post_init behavior." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Tensor-parallelism is the only overlap; one is a general training bug, the other is a specific OOM on loading a Qwen model." + "reason": "Both mention tensor parallelism, but one is a generic training bug and the other is an OOM while loading a specific Qwen2 model; not the same bug." }, { "left": "issue:43867", "right": "issue:44484", "accept": false, - "reason": "Load error from sorted state_dict is unrelated to the save_pretrained shard-size question." + "reason": "Loading a sorted state_dict and save_pretrained max_shard_size semantics are unrelated." }, { "left": "issue:43618", "right": "issue:44857", "accept": false, - "reason": "Different model areas and different failure modes." + "reason": "CLIP attentions regression and float16 AMP crash in LwDetrImageLoss are unrelated." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "ImportError for missing PyTorch vs tokenizer-loading regression; unrelated." + "reason": "Different subsystems: missing PyTorch dependency for AutoModel import vs tokenizer loading regression." }, { "left": "issue:43618", "right": "issue:44977", "accept": false, - "reason": "Different models and different runtime paths; flash-attention generation bug is not the same issue." + "reason": "CLIPOutput attentions issue is unrelated to Qwen3.5 flash-attention generation failure." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "Test seeding flakiness vs a specific multimodal model loading failure." + "reason": "Missing set_seed in tests is unrelated to loading microsoft/Phi-4-multimodal-instruct." }, { "left": "issue:42890", "right": "issue:44112", "accept": false, - "reason": "Different CI/test issues with different symptoms and files." + "reason": "Both are CI/test-related, but one is set_seed coverage and the other is a stale device-override test failure; not the same bug." }, { "left": "issue:43010", "right": "issue:44075", "accept": false, - "reason": "No concrete overlap beyond both being optimizer/model-related bugs." + "reason": "Different code paths: cache update no_grad decoration vs optimizer SGD args being ignored." }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Both are loading regressions, but one is tokenizer-loading and the other is checkpoint reload after version upgrade; not the same bug." + "reason": "Both concern loading, but one is tokenizer loading in v5 and the other is reloading checkpoints after upgrade; too broad and not the same concrete defect." }, { "left": "issue:42890", "right": "issue:45341", "accept": false, - "reason": "Different test problems; one is seed-related, the other is a separate testing_utils bug." + "reason": "Unrelated test issues: set_seed omissions vs a bug in testing_utils.py." }, { "left": "issue:43502", "right": "issue:44062", "accept": false, - "reason": "Unrelated API-request behavior vs AddedToken argument handling." + "reason": "Different problems: local_files_only still triggering API requests vs tokenizers.AddedToken() receiving duplicate 'special' kwargs." }, { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Memory regression under int4 quantization is a different failure mode from checkpoint reload incompatibility." + "reason": "OOM under int4 quantization is unrelated to checkpoint reload failures after upgrading from v4.57 to v5.x." }, { "left": "issue:44222", "right": "issue:45125", "accept": false, - "reason": "Different Qwen3.5 MoE issues: FP8 save_pretrained vs missing tensor-parallel plan." + "reason": "Different MoE/tensor-parallel topics: FP8 save_pretrained versus missing _tp_plan for tensor parallelism." }, { "left": "issue:43010", "right": "issue:43065", "accept": false, - "reason": "No meaningful overlap: cache/layer update decoration vs dummy Conv2d in Sam3PixelDecoder." + "reason": "No overlap: no_grad decoration for cache updates vs a dummy Conv2d in Sam3PixelDecoder." }, { "left": "issue:43010", "right": "issue:43232", "accept": false, - "reason": "Different generation-path bugs with no shared concrete code-path shown." + "reason": "Different generation/cache bugs; the titles point to separate code paths and symptoms." }, { "left": "issue:45125", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 MoE, but one is tensor-parallel-plan metadata and the other is a generic from_pretrained failure; too broad to merge." + "reason": "The Qwen3.5 MoE from_pretrained error may be related to tensor parallelism, but there is no evidence it is the same concrete bug as the missing _tp_plan issue." }, { "left": "issue:42548", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different symbols and likely different export problems." + "reason": "Both are import-name errors, but they involve different exported symbols (PreTrainedModel vs HybridCache) and likely different causes." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Both are Qwen3.5 MoE vision-related, but one is TP metadata and the other is a missing config field." + "reason": "Both mention Qwen3.5 MoE-related config/loading concerns, but one is missing _tp_plan and the other is a dropped vision config field; different defects." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Running compiled files vs slower filtering after upgrade are unrelated." + "reason": "Compiled Python file errors and a slowdown after upgrading Transformers are unrelated." }, { - "left": "issue:30990", - "right": "issue:31795", + "left": "issue:43502", + "right": "issue:44485", "accept": false, - "reason": "SentenceTransformer loading hang vs docstring confusion; no shared bug." + "reason": "local_files_only API leakage and max_shard_size defaults are unrelated save/load behaviors." }, { - "left": "issue:43502", - "right": "issue:44485", + "left": "issue:30990", + "right": "issue:31795", "accept": false, - "reason": "Local-files-only network leakage is unrelated to RoPE implementation discussion." + "reason": "Sentence Transformers loading hang and a docstring clarification for forward args are not the same bug." }, { "left": "issue:44075", "right": "issue:44871", "accept": false, - "reason": "Optimizer argument usage bug vs eos_token_id config mismatch; unrelated." + "reason": "Optimizer argument handling and Gemma eos_token_id config mismatch are unrelated." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Different subsystems and different model behaviors." + "reason": "Different topics entirely: Liger Kernel application in model_init vs a DSA indexer missing ReLU." + }, + { + "left": "issue:43010", + "right": "issue:44368", + "accept": false, + "reason": "Cache update no_grad and a warning about tie_word_embeddings=False are unrelated." }, { "left": "issue:42890", "right": "issue:44373", "accept": false, - "reason": "CI/test flakiness vs a docstring correction; not the same issue." + "reason": "Test seed flakiness and a wrong docstring for position_ids are unrelated." }, { "left": "issue:43502", "right": "issue:44222", "accept": false, - "reason": "Different features and failure modes; no common underlying bug." + "reason": "API requests despite local_files_only and FP8 save_pretrained for MoE are unrelated." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Both are version-related loading problems, but one is model-specific and the other is a general checkpoint reload regression." + "reason": "Qwen3VL MoE loading break in dev0 and generic checkpoint reload failures after upgrading are similar only at a high level, not the same concrete issue." } ] }, @@ -12974,14 +12980,14 @@ ], "soft_pairs": [ "issue:43502|issue:44961", - "issue:44355|issue:44655", "issue:44222|issue:44230", + "issue:44355|issue:44655", "issue:44355|issue:45071", "issue:42831|issue:43874", "issue:33453|issue:34689", "issue:42757|issue:43576", - "issue:33453|issue:36296", "issue:43827|issue:44961", + "issue:33453|issue:36296", "issue:43576|issue:45071", "issue:43576|issue:43824", "issue:43673|issue:44492", @@ -12996,8 +13002,8 @@ "issue:44485|issue:44961", "issue:36296|issue:43526", "issue:36296|issue:43012", - "issue:44273|issue:45003", "issue:41628|issue:44355", + "issue:44273|issue:45003", "issue:44261|issue:44360", "issue:44273|issue:45542", "issue:44623|issue:45230", @@ -13011,136 +13017,136 @@ "issue:44704|issue:44908", "issue:44246|issue:44273" ], - "prepared_review_unit_hash": "bf848d6c42b5257b86bbd5b4886101ff84e9cab628094d63fbcb71d31c809f64", + "prepared_review_unit_hash": "aae42b06cd365e0f357b9a57d63b6fdee80eb4e2200ddc1a6a471be399d6fe7f", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 37, - "serialized_chars": 37413, - "estimated_input_tokens": 9354, - "estimated_eval_tokens": 18964 + "serialized_chars": 37410, + "estimated_input_tokens": 9353, + "estimated_eval_tokens": 18962 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly loose lexical similarity around Transformers v5/import/loading/cache issues, not true duplicates. I wouldn\u2019t merge any soft-edge pairs; they span different code paths, models, and failure modes. No PRs are present.", - "confidence": 0.79, - "canonical_issue_reason": "No single issue is a clear duplicate anchor for the whole cluster. If one must be chosen, issue 41669 is the broadest/system-level representative because it concerns package import behavior that can affect many downstream reports.", + "summary": "The items are largely unrelated regression/bug reports across different subsystems. No soft pair looks like the same underlying bug or change, so all proposed edges are rejected. The strongest standalone issue to anchor triage is the broad import-performance report.", + "confidence": 0.84, + "canonical_issue_reason": "If one issue must represent this set, issue 44246 is the best anchor because it is broad, user-visible, and actionable (import performance regression), unlike the more narrowly scoped or highly specialized reports.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 41669 is the best representative only in a broad triage sense: it is the most systemic and central to the import/export side of the cluster, though the cluster is too heterogeneous to treat it as a true canonical duplicate.", + "best_issue_reason": "Issue 44246 is the most generally applicable and triage-friendly: it affects core import behavior, has a clear symptom, and is not tied to a single niche model or feature path.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43502", "right": "issue:44961", "accept": false, - "reason": "Unrelated: local_files_only/network access bug vs an unrelated placeholder issue." + "reason": "Unrelated: local_files_only network leakage vs a placeholder/title-only issue." }, { - "left": "issue:44355", - "right": "issue:44655", + "left": "issue:44222", + "right": "issue:44230", "accept": false, - "reason": "Different failures: compiled Python file errors vs saving Pipeline objects." + "reason": "Both are FP8/MoE-related, but save_pretrained persistence and inference support are different code paths." }, { - "left": "issue:44222", - "right": "issue:44230", + "left": "issue:44355", + "right": "issue:44655", "accept": false, - "reason": "Both involve FP8/MoE, but one is save_pretrained and the other is inference support; not the same bug." + "reason": "Different failures: running compiled Python files vs saving Pipeline objects." }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Different code paths: compiled-file runtime errors vs PretrainedConfig type checking regression." + "reason": "Different bugs: compiled-file execution errors vs PretrainedConfig type-checking regression." }, { "left": "issue:42831", "right": "issue:43874", "accept": false, - "reason": "Accuracy issue in FineGrainedFP8 is unrelated to missing image-patch counting in a processor." + "reason": "Different subsystems: FP8 accuracy vs missing image-processor method." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Both are loading regressions, but for different models and likely different root causes." + "reason": "Both are loading regressions, but they involve different models and different failure modes." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "Hugging Face Hub import error vs transformers env CLI breakage; too different." + "reason": "Different issues: missing huggingface_hub import vs broken transformers env command." }, { - "left": "issue:33453", - "right": "issue:36296", + "left": "issue:43827", + "right": "issue:44961", "accept": false, - "reason": "Tokenizer/model loading regression vs tensor parallel training bug; unrelated." + "reason": "Docs referencing pipeline() is unrelated to the placeholder issue." }, { - "left": "issue:43827", - "right": "issue:44961", + "left": "issue:33453", + "right": "issue:36296", "accept": false, - "reason": "Docs still referencing pipeline() is unrelated to the placeholder issue." + "reason": "Tokenizer loading regression and tensor-parallel training bug are unrelated." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "Both are v5 regressions, but CLI/env behavior and PretrainedConfig typing are distinct problems." + "reason": "Both are v5 regressions, but command behavior and type checking are separate problems." }, { "left": "issue:43576", "right": "issue:43824", "accept": false, - "reason": "Different import/CLI issues with no shared concrete failure path shown." + "reason": "Broken env command and missing Qwen2.5-VL export are unrelated." }, { "left": "issue:43673", "right": "issue:44492", "accept": false, - "reason": "Cache missing in chunked_prefill is not the same as a typo in cache strategies." + "reason": "Cache-missing runtime bug vs a typo in cache strategy text; not the same issue." }, { "left": "issue:38617", "right": "issue:43576", "accept": false, - "reason": "ImportError for layer_type_validation is different from the env command regression." + "reason": "ImportError for layer_type_validation is unrelated to the env command regression." }, { "left": "issue:33357", "right": "issue:33453", "accept": false, - "reason": "MacOS bus error with CLIP model is unrelated to tokenizer loading regression." + "reason": "Different regressions affecting different loading paths and models." }, { "left": "issue:44961", "right": "issue:45468", "accept": false, - "reason": "Placeholder issue is unrelated to Gemma audio positional encoding." + "reason": "Placeholder issue is unrelated to Gemma-4 audio positional encoding." }, { "left": "issue:31795", "right": "issue:38617", "accept": false, - "reason": "Documentation confusion is unrelated to the import error." + "reason": "Documentation confusion vs an ImportError are not the same underlying bug." }, { "left": "issue:43576", "right": "issue:44351", "accept": false, - "reason": "CLI env breakage and HybridCache import failure are separate issues." + "reason": "Broken env command and missing HybridCache import are separate breakages." }, { "left": "issue:44908", @@ -13152,103 +13158,115 @@ "left": "issue:41628", "right": "issue:42548", "accept": false, - "reason": "Both are import errors, but for different symbols and likely different broken exports." + "reason": "Both are import errors, but for different symbols and likely different re-export paths." }, { "left": "issue:36296", "right": "issue:43901", "accept": false, - "reason": "Tensor parallel training bug is unrelated to a docs/behavior mismatch in TextClassificationPipeline." + "reason": "Tensor-parallel training bug vs pipeline docs mismatch are unrelated." }, { "left": "issue:44485", - "right": "issue:45071", + "right": "issue:44961", "accept": false, - "reason": "RoPE implementation discussion is unrelated to PretrainedConfig type checking." + "reason": "GLM-5 RoPE implementation discussion is unrelated to the placeholder issue." }, { "left": "issue:36296", "right": "issue:43526", "accept": false, - "reason": "Tensor parallel training and BeitImageProcessorFast reduce_labels are different bugs." + "reason": "Tensor-parallel training and BeitImageProcessorFast label reduction are different bugs." }, { "left": "issue:36296", "right": "issue:43012", "accept": false, - "reason": "Tensor parallel training bug is unrelated to a bfloat16 compilation warning." + "reason": "Training bug vs bfloat16 compile warning are unrelated." }, { - "left": "issue:44273", - "right": "issue:45003", + "left": "issue:41628", + "right": "issue:44355", "accept": false, - "reason": "Lazy loading behavior and unsafe sys.modules access are related in theme but not the same concrete bug." + "reason": "Different import/export failure and compiled-file execution problem." }, { - "left": "issue:41628", - "right": "issue:44355", + "left": "issue:44273", + "right": "issue:45003", "accept": false, - "reason": "Import error and compiled-file runtime errors are different failure modes." + "reason": "Lazy loading symptoms may sound related, but the titles describe different failure modes and code paths." }, { "left": "issue:44261", "right": "issue:44360", "accept": false, - "reason": "MLA layernorm precision issue is unrelated to the DSA indexer ReLU discussion." + "reason": "Precision error in MLA q_a_layernorm is unrelated to the DSA indexer/ReLU discussion." }, { "left": "issue:44273", "right": "issue:45542", "accept": false, - "reason": "Lazy loading bug is unrelated to the TensorFlow backend installation error." + "reason": "Lazy loading issue and TensorFlow backend error are unrelated." }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "Missing files in processor.save_pretrained is unrelated to a generic bug report placeholder." + "reason": "Missing processor.save_pretrained files is not the same as a generic bug report." }, { "left": "issue:43576", "right": "issue:44556", "accept": false, - "reason": "Both mention v5 breakage, but CLI env behavior and checkpoint reload incompatibility are different issues." + "reason": "Both are v5-related breakages, but a broken CLI/env command is not the same as checkpoint reload incompatibility." }, { "left": "issue:41669", "right": "issue:43673", "accept": false, - "reason": "Import-star cleanup/import performance is unrelated to cache missing during chunked_prefill." + "reason": "Import-star cleanup/performance issue is unrelated to GenerationMixin cache behavior." }, { "left": "issue:44393", "right": "issue:45071", "accept": false, - "reason": "Qwen3-VL bbox hallucination/output error is unrelated to PretrainedConfig type checking." + "reason": "Qwen3-VL bbox hallucination/error and PretrainedConfig type checking are unrelated." }, { "left": "issue:44038", "right": "issue:45125", "accept": false, - "reason": "Related family, but not the same concrete bug: general Qwen3-VL-Moe issue vs missing _tp_plan for tensor parallelism." + "reason": "General Qwen3-VL-Moe bug and missing _tp_plan for tensor parallelism are different concrete issues." }, { "left": "issue:44556", "right": "issue:45092", "accept": false, - "reason": "Both are checkpoint compatibility problems, but one is generic reload regression and the other is remote-code/meta-init incompatibility." + "reason": "Both involve v5 compatibility, but they concern different checkpoint/model incompatibility causes." }, { "left": "issue:42371", "right": "issue:44492", "accept": false, - "reason": "TF32 API guidance and a cache-strategies typo are unrelated." + "reason": "TF32 behavior settings and a cache-strategies typo are unrelated." }, { "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "Flaky test seeding issue is unrelated to inverse_sqrt scheduler kwargs handling." + "reason": "Test flakiness from missing set_seed() is unrelated to scheduler kwargs handling." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor cached_file kwargs passthrough and inverse_sqrt scheduler kwargs are different bugs." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import performance regression and lazy-loading malfunction are related in theme but not clearly the same underlying bug." } ] }, @@ -13278,13 +13296,13 @@ "issue:37428", "issue:41669", "issue:42371", - "issue:42890", "issue:42913", "issue:43066", "issue:43352", "issue:43408", "issue:43502", "issue:43519", + "issue:43525", "issue:43576", "issue:43606", "issue:43618", @@ -13360,284 +13378,324 @@ "issue:44263|issue:45468", "issue:44261|issue:45468", "issue:43606|issue:45468", - "issue:42890|issue:43704", + "issue:43704|issue:45341", "issue:42913|issue:43066", "issue:44908|issue:44998", "issue:44861|issue:45290", "issue:44998|issue:45341", "issue:44492|issue:44998", "issue:43618|issue:44998", - "issue:44556|issue:44998" + "issue:44556|issue:44998", + "issue:43525|issue:44861" ], - "prepared_review_unit_hash": "d779147d46acadceba8f2358d08b25486ecb65d0f83b63cfd6f16f0e17f88bcb", + "prepared_review_unit_hash": "11e42a2b65ae3e14f8046cd1449bbbc79b34adc4ddafdbb3dee411ee01843896", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 46, - "serialized_chars": 40860, - "estimated_input_tokens": 10215, - "estimated_eval_tokens": 20686 + "soft_pair_count": 47, + "serialized_chars": 41102, + "estimated_input_tokens": 10276, + "estimated_eval_tokens": 20808 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is a broad mix of v5 regressions, tokenizer/loading bugs, model-specific failures, and unrelated docs/test/config issues. I don\u2019t see any exact duplicate pair among the soft edges; most are false positives from vague similarity.", - "confidence": 0.89, - "canonical_issue_reason": null, + "summary": "Most pairs are clearly unrelated; the only plausible duplicate is the v5 tokenizer behavior pair (42913/43066). I would keep the specific decoder-type regression as the main issue target.", + "confidence": 0.81, + "canonical_issue_reason": "issue:43066 is the most concrete tokenizer/v5 regression here: it names the specific wrong decoder-type behavior, making it the clearest duplicate target over the broader symptom report.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43066 is the clearest concrete tokenizer-v5 regression here and a better representative than the many generic or unrelated items, but the cluster is too heterogeneous for a true canonical duplicate.", + "best_issue_reason": "issue:43066 is the strongest standalone issue in this set because it is specific, reproducible, and actionable; the broader tokenizer-behavior report fits beneath it.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "Different model/type-mismatch warning vs. Liger Kernel application behavior; not the same bug." + "reason": "Different problems: model-type mismatch warning vs Liger Kernel application in model_init." }, { "left": "issue:44297", "right": "issue:45230", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch vs. an unspecified bug report; no shared concrete failure." + "reason": "Tokenizer save_pretrained bug vs a generic bug report; no clear same underlying failure." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "Broken env CLI in v5 vs. inverse_sqrt scheduler kwarg handling; unrelated subsystems." + "reason": "Unrelated: broken env command vs inverse_sqrt scheduler kwargs handling." }, { "left": "issue:43352", "right": "issue:44393", "accept": false, - "reason": "Flash Attention support error for Nemotron vs. Qwen3-VL bbox output issue; different models and failures." + "reason": "Flash Attention 2.0 support error vs Qwen3-VL bbox hallucination; different code paths." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python version compatibility regression vs. Qwen3-VL output bug; not the same change." + "reason": "Python compatibility regression vs Qwen3-VL bounding-box output bug." }, { "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwarg bug vs. PretrainedConfig type checking breakage; different code paths." + "reason": "Scheduler kwargs ignored vs PretrainedConfig type-checking breakage." }, { "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "position_ids buffer read error vs. generic bug report; insufficient overlap." + "reason": "Position_ids buffer access issue vs generic bug report; insufficient evidence they are the same." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Import * cleanup/performance issue vs. HybridCache import error; related to imports only at a very broad level." + "reason": "Import-star cleanup/perf issue is unrelated to HybridCache import failure." }, { "left": "issue:41669", "right": "issue:45341", "accept": false, - "reason": "Model import cleanup vs. testing_utils bug; unrelated." + "reason": "Different targets: model import cleanup vs testing_utils bug." }, { "left": "issue:44492", "right": "issue:44945", "accept": false, - "reason": "Cache-strategy typo vs. pipeline-parallelism output bug; different problems." + "reason": "Cache-strategy typo is unrelated to pipeline-parallelism output correctness." }, { "left": "issue:41669", "right": "issue:44492", "accept": false, - "reason": "Import hygiene/perf issue vs. a cache-strategy typo; not the same underlying change." + "reason": "Broad import cleanup does not match a cache-strategy typo bug." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Liger Kernel application with model_init vs. flash_attention_3 degenerate training; different mechanisms." + "reason": "Liger Kernel not applied in model_init is different from flash_attention_3 training degeneration." }, { "left": "issue:44393", "right": "issue:45478", "accept": false, - "reason": "Qwen3-VL 2D bbox issue vs. Qwen3.5 MoE from_pretrained error; same family name, different bugs." + "reason": "Different models and failures: Qwen3-VL bbox output vs Qwen3.5 MoE from_pretrained error." }, { "left": "issue:44077", "right": "issue:44861", "accept": false, - "reason": "patchtsmixer post_init policy vs. tied-weights key handling crash; unrelated." + "reason": "PatchTSMixer post_init policy issue vs tied-weight key AttributeError." }, { "left": "issue:43827", "right": "issue:44908", "accept": false, - "reason": "Docs still referencing pipeline() vs. inverse_sqrt scheduler kwargs; different areas." + "reason": "Docs still referencing pipeline() vs scheduler kwargs bug; not the same change." }, { "left": "issue:31795", "right": "issue:33453", "accept": false, - "reason": "Forward-doc confusion vs. tokenizer loading regression; no shared bug." + "reason": "Documentation clarification for model.forward is unrelated to tokenizer loading regression." }, { "left": "issue:41669", "right": "issue:44908", "accept": false, - "reason": "Import cleanup in models vs. scheduler kwarg handling; unrelated." + "reason": "Import-star cleanup/perf issue is unrelated to scheduler behavior." }, { "left": "issue:43723", "right": "issue:44393", "accept": false, - "reason": "Tokenizer loading in v5 vs. Qwen3-VL bounding-box output; different failures." + "reason": "Tokenizer loading failure vs Qwen3-VL bounding-box bug; different subsystems." }, { "left": "issue:44655", "right": "issue:44861", "accept": false, - "reason": "Pipeline save_pretrained failure vs. tied-weight key crash; distinct save/load paths." + "reason": "Saving Pipeline objects is unrelated to tied-weight key AttributeError." }, { "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "CLI env command regression vs. lazy loading bug; unrelated." + "reason": "Broken env command and lazy loading bug are unrelated." }, { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "DSA indexer ReLU concern vs. Gemma-4 audio positional encoding issue; different code paths." + "reason": "DSA indexer ReLU discussion vs Gemma4Audio positional encoding; not the same issue." }, { "left": "issue:45230", "right": "issue:45310", "accept": false, - "reason": "Generic bug report vs. Qwen3.5 MoE from_pretrained error; no evidence they are the same issue." + "reason": "Generic bug report does not establish the same root cause as the Qwen3.5 MoE from_pretrained error." }, { "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Lazy loading bug vs. scheduler kwarg bug; unrelated." + "reason": "Lazy loading problem vs scheduler kwargs bug." }, { "left": "issue:41669", "right": "issue:43704", "accept": false, - "reason": "Import * cleanup/perf issue vs. VRAM leak in dataloader threads; different bugs." + "reason": "Import-star usage/performance issue is unrelated to VRAM leak across dataloader threads." }, { "left": "issue:43519", "right": "issue:44485", "accept": false, - "reason": "Qwen3VL timestamp calculation vs. GLM-5 RoPE implementation; different model logic." + "reason": "Timestamp calculation bug vs GLM-5 RoPE implementation issue." }, { "left": "issue:43519", "right": "issue:44908", "accept": false, - "reason": "Timestamp calculation bug vs. scheduler kwargs bug; unrelated." + "reason": "Qwen3VL timestamp bug is unrelated to scheduler kwargs handling." }, { "left": "issue:31795", "right": "issue:34689", "accept": false, - "reason": "Documentation confusion vs. Llama 3.2 model loading regression; not the same underlying problem." + "reason": "Model.forward docs issue vs Llama 3.2 vision model loading break." }, { "left": "issue:31795", "right": "issue:37428", "accept": false, - "reason": "Docs issue vs. flash-attention import error; unrelated." + "reason": "Documentation confusion vs flash_attention_utils import error." }, { "left": "issue:44315", "right": "issue:45092", "accept": false, - "reason": "Liger Kernel/model_init behavior vs. meta-init checkpoint incompatibility; related to initialization broadly, but not the same bug." + "reason": "Liger Kernel/model_init issue is different from old checkpoint meta-init incompatibility." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "VRAM leak in multithreaded use vs. RoPE implementation discussion; different problems." + "reason": "VRAM leak across threads is unrelated to RoPE implementation." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "Different architectures and failures: GLM-5 RoPE vs Gemma4 audio positional encoding." }, { "left": "issue:42371", "right": "issue:43704", "accept": false, - "reason": "TF32 behavior settings vs. VRAM leak; unrelated." + "reason": "TF32 API-setting guidance is unrelated to VRAM leak behavior." }, { "left": "issue:43502", "right": "issue:43519", "accept": false, - "reason": "local_files_only still triggers API requests vs. timestamp calculation in a processor; different issues." + "reason": "local_files_only API requests vs timestamp calculation bug; different paths." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Same as above: unrelated documentation issue vs model-loading regression." + }, + { + "left": "issue:31795", + "right": "issue:37428", + "accept": false, + "reason": "No shared underlying bug: docs wording vs import failure." + }, + { + "left": "issue:44315", + "right": "issue:45092", + "accept": false, + "reason": "Both involve loading, but one is Liger Kernel application and the other is checkpoint/meta-init compatibility; not the same bug." + }, + { + "left": "issue:43704", + "right": "issue:45341", + "accept": false, + "reason": "VRAM leak issue is unrelated to testing_utils bug." }, { "left": "issue:42913", "right": "issue:43066", - "accept": false, - "reason": "Both touch tokenizer-v5 behavior, but one is generic behavior drift and the other is a specific decoder-type bug; not enough to treat as the same change." + "accept": true, + "reason": "Both describe the same v5 tokenizer regression; 43066 is the concrete wrong-decoder-type manifestation of the broader behavior change in 42913." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwarg bug vs. unrelated noise issue." + "reason": "Scheduler kwargs bug is unrelated to the placeholder 'Unemployment' issue." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied-weight key crash vs. chat template crash with tool calls; different code paths." + "reason": "Tied-weight key crash and apply_chat_template tool-call crash are different failures." }, { "left": "issue:44998", "right": "issue:45341", "accept": false, - "reason": "Unemployment joke issue vs. testing_utils bug; unrelated." + "reason": "Placeholder issue title is unrelated to testing_utils." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "Cache-strategy typo vs. unrelated noise issue." + "reason": "Cache-strategy typo does not match the placeholder issue." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions regression vs. unrelated noise issue." + "reason": "CLIPOutput attentions regression is unrelated to the placeholder issue." }, { "left": "issue:44556", "right": "issue:44998", "accept": false, - "reason": "Checkpoint reload regression after v4.57->v5.x upgrade vs. unrelated noise issue." + "reason": "Checkpoint reload incompatibility is unrelated to the placeholder issue." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Llama4Config missing pad_token_id vs tied-weight key AttributeError; different code paths." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is conservative and grounded in the packet. The only accepted soft pair, issue:42913 / issue:43066, is supported by the titles as the specific v5 tokenizer regression versus the broader tokenizer-behavior report. The remaining pair judgments stay on the cautious side and do not overclaim shared root cause from the available evidence." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -13661,7 +13719,6 @@ "issue:36683", "issue:38617", "issue:43298", - "issue:43525", "issue:43572", "issue:44230", "issue:44623", @@ -13669,7 +13726,6 @@ "issue:45310" ], "soft_pairs": [ - "issue:43525|issue:44861", "issue:43572|issue:44861", "issue:31515|issue:38617", "issue:44230|issue:44623", @@ -13680,94 +13736,88 @@ "issue:31515|issue:36683", "issue:43298|issue:44230" ], - "prepared_review_unit_hash": "d4b4f71694ad9f5dbd00d15223ede4d36e8fbde4324a70be08cc61fe1ecbc969", + "prepared_review_unit_hash": "71ff70509adff5f4d7c4fb99dfc5e35ca9465de20eb5b7ae0110da197c9e055c", "component_budget": { "node_count": 366, "item_count": 366, "soft_pair_count": 1367, - "serialized_chars": 705290, - "estimated_input_tokens": 176323, - "estimated_eval_tokens": 352902 + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 }, "budget": { - "node_count": 13, - "item_count": 13, - "soft_pair_count": 10, - "serialized_chars": 10324, - "estimated_input_tokens": 2581, - "estimated_eval_tokens": 5418 + "node_count": 12, + "item_count": 12, + "soft_pair_count": 9, + "serialized_chars": 9455, + "estimated_input_tokens": 2364, + "estimated_eval_tokens": 4984 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is only loosely related: most items mention model loading/configuration or recent regressions, but they cover distinct bugs, docs, and feature gaps across different code paths. No soft pair looks like a true duplicate.", - "confidence": 0.34, - "canonical_issue_reason": "Issue 45310 is the most concrete and representative load-time regression in the set: it has a specific from_pretrained failure, a clear model family, and an explicit version boundary. The rest of the cluster is too heterogeneous to serve as one duplicate anchor.", + "summary": "The soft edges look like false positives: the paired issues cover unrelated docs, loading, config, tokenizer, fp8, and training problems. No duplicate or merge-worthy issue pair stands out.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "45310 is the best single issue to anchor the cluster because it is recent, specific, and directly about a reproducible from_pretrained error rather than a broad feature request or documentation problem.", + "best_issue_reason": "issue:45310 is the strongest standalone report: it is specific, reproducible, and names a concrete from_pretrained regression tied to transformers>=5.4.0 and Qwen3.5 MoE.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43525", - "right": "issue:44861", - "accept": false, - "reason": "Different bugs: Llama4Config missing pad_token_id versus _get_tied_weight_keys crashing on a list-typed _tied_weights_keys. Separate code paths and fixes." - }, { "left": "issue:43572", "right": "issue:44861", "accept": false, - "reason": "StableLmConfig pad_token_idx regression and tied-weight-key helper crash are unrelated failure modes, not the same underlying bug." + "reason": "Different failures: StableLmConfig pad_token_idx regression vs AttributeError in tied-weight key handling. Not the same bug." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Checkpoint loading is slow versus an ImportError from configuration_utils; same broad area, but not the same defect." + "reason": "Checkpoint loading performance issue vs ImportError for layer_type_validation. Different code paths and symptoms." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "fp8 inference support request and processor.save_pretrained missing files are different features/issues with no shared code-path bug." + "reason": "FP8 inference support request for Qwen models vs processor.save_pretrained missing files. Unrelated issues." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "Slow from_pretrained checkpoint loading and tensor parallel training bug affect different subsystems and symptoms." + "reason": "from_pretrained slowdown vs tensor-parallel training bug. Not the same underlying problem." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "One is a loading performance issue, the other is documentation confusion about model.forward arguments." + "reason": "Slow checkpoint loading vs documentation confusion about input_ids/past_key_values. Different scope entirely." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "Both mention Qwen3.5 MoE/fp8 loading, but one is a feature/support request while the other is a concrete from_pretrained regression; not the same bug." + "reason": "Both mention Qwen3.5 MoE, but one is FP8 inference support and the other is a from_pretrained runtime error. Not mergeable as one bug." }, { "left": "issue:31515", "right": "issue:33453", "accept": false, - "reason": "Checkpoint load slowness and tokenizer loading regression are both loading-related, but they are distinct bugs with different symptoms." + "reason": "Checkpoint loading slowness vs tokenizer loading regression. Separate loading layers and failure modes." }, { "left": "issue:31515", "right": "issue:36683", "accept": false, - "reason": "from_pretrained slowness is unrelated to Gemma3Config missing vocab_size, which is an attribute/config regression." + "reason": "General loading slowdown vs Gemma3Config missing vocab_size. Different model/config bug." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Continuous batching not supporting audio models and fp8 support for Qwen3.5/Qwen3-VL are different capability gaps." + "reason": "Continuous batching audio-model limitation vs FP8 support for Qwen3.5. Different features and code paths." } ] }, @@ -13797,12 +13847,12 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "51515c94aa8d6b22e844001107f04c8248c8691509689266725b8cf92739b3ab", + "prepared_review_unit_hash": "3c4d23a14b9ac86b9d8543a753ef90f87db16047fea82ca136ecc9c6d5512e5e", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1589, + "serialized_chars": 1590, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -13810,7 +13860,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1589, + "serialized_chars": 1590, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -13819,18 +13869,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate community contribution feature requests: one for Flash Attention 2 support across more architectures, and one for GGUF support across more architectures. They are thematically similar but not the same underlying change.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "The two issues are both community-contribution feature requests, but they target different features and code paths: Flash Attention 2 support vs GGUF support. They are not duplicates.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 26350 is the stronger representative of the cluster by discussion activity, but the two issues are not the same request, so there is no true canonical duplicate here.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 26350 has substantially more discussion activity and is a clearer, higher-signal feature request than 33260.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different requested capabilities and code paths: Flash Attention 2 vs GGUF. Similar broad theme, but not a duplicate bug/change." + "reason": "Different requested features for different backend/architecture support; they do not describe the same underlying bug or change." } ] }, @@ -13867,7 +13917,7 @@ "issue:43834|issue:44365", "issue:44365|issue:44370" ], - "prepared_review_unit_hash": "bd09e4f50634946b0d23c2281a88c76db7b6d0b40cb60d5c29023b482731ff79", + "prepared_review_unit_hash": "3f549b441ca0c48ab82b5271d4b6c621b11270bf9cc00fe7cb5fd1be570a63d9", "component_budget": { "node_count": 4, "item_count": 4, @@ -13889,48 +13939,48 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These look like near-duplicate translation-request templates, but only issue 27088 is concretely about Turkish. The other three are redacted/generic placeholders, so I would not collapse them without fuller bodies.", - "confidence": 0.61, - "canonical_issue_reason": "Issue 27088 is the only one with a specific language target (Turkish) and it is the oldest/open item, making it the clearest representative of this cluster.", + "summary": "The cluster is dominated by nearly identical i18n translation-request templates, but the only concrete item is Turkish. These look like separate language requests rather than duplicates of the same underlying issue, so the high textual similarity is misleading.", + "confidence": 0.89, + "canonical_issue_reason": "issue #27088 is the only specific, actionable request (Turkish) and is the oldest open item; the others are generic placeholder-style translation issues.", "canonical_pr_reason": null, - "best_issue_reason": "27088 is the most concrete and actionable issue; the others read like generic template instances with no explicit language evidence.", + "best_issue_reason": "#27088 is the best representative because it names a real language and can actually be acted on, while the others are template placeholders with no concrete language filled in.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Very similar translation-template wording, but only the left issue names Turkish; the right issue is generic and could be a different language request." + "reason": "Both are translation-request issues, but they appear to be for different language entries / template instances, not the same concrete change or bug." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Same template pattern, but the specific language on the right is unknown here, so I can't confirm the same underlying request." + "reason": "The shared wording is just an i18n template; #27088 is specifically Turkish, so this is not the same underlying issue." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "The left is explicitly Turkish while the right is a generic placeholder; that is not enough to treat them as the same issue." + "reason": "High textual similarity comes from the template, but there is no evidence they target the same language or same work item." }, { "left": "issue:43834", "right": "issue:44370", "accept": false, - "reason": "Both are generic translation-template issues, but the available data does not prove they target the same language or same request." + "reason": "These are near-identical templates, but without a shared concrete language target they should not be merged as duplicates." }, { "left": "issue:43834", "right": "issue:44365", "accept": false, - "reason": "Highly similar template text, but no concrete language/body targets are available to establish a true duplicate." + "reason": "The bodies/titles are template-like and likely represent separate language requests rather than one duplicate issue." }, { "left": "issue:44365", "right": "issue:44370", "accept": false, - "reason": "Near-identical placeholder titles alone are insufficient; they may be separate language translation requests rather than duplicates." + "reason": "Same template family, but no concrete evidence of the same underlying translation task; treat as separate issues." } ] }, @@ -13995,184 +14045,178 @@ "issue:43837|issue:45027", "issue:43837|issue:45177" ], - "prepared_review_unit_hash": "a77aaabff727343b0a0bb9264db6060026a38b16502e2f2bb86b3a06d788c5ca", + "prepared_review_unit_hash": "a44b4221c633378924b1ce3284565b95f866975bc4b31b9cd2c4078abdd028df", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 25, - "serialized_chars": 16010, - "estimated_input_tokens": 4003, - "estimated_eval_tokens": 8262 + "serialized_chars": 16008, + "estimated_input_tokens": 4002, + "estimated_eval_tokens": 8260 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 25, - "serialized_chars": 16010, - "estimated_input_tokens": 4003, - "estimated_eval_tokens": 8262 + "serialized_chars": 16008, + "estimated_input_tokens": 4002, + "estimated_eval_tokens": 8260 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are mostly unrelated model-request issues that only share generic wording like \u201cadd support\u201d or \u201cproposal to add.\u201d They span different model families and modalities (BERT variants, audio/TTS/ASR, OCR, vision, sparsity), so they are not true duplicates.", - "confidence": 0.87, - "canonical_issue_reason": "issue:43837 is the most central hub in the similarity graph, but the cluster is still heterogeneous; none of these appear to be the same underlying request.", + "summary": "This cluster is a loose set of model-add requests spanning different families and modalities (text BERT variants, speech ASR/TTS, audio, OCR, and one sparse-transformer proposal). They are not the same underlying bug/change, so none of the soft links should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43837 is the best representative of the cluster because it has the most soft-similarity links, though it is not a true duplicate of the others.", + "best_issue_reason": "issue:37015 ('Add NeoBERT') is the most generic and actively discussed model-add request in the set, so it is the best loose representative of the cluster\u2014but it is not a true duplicate target for the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different model families: EuroBERT vs DEIMv2. Shared \u201cadd model\u201d framing is too generic." + "reason": "Different model requests: EuroBert vs DEIMv2 are unrelated model families and likely different implementation paths." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBERT and Qwen3-TTS are unrelated model additions with different modalities." + "reason": "Text model config request vs Qwen3-TTS speech support; different modality and code path." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are different models and different tasks." + "reason": "NeoBERT and Qwen3-ASR are different models for different tasks; not the same change." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT vs weight-sparse transformer/circuitgpt are distinct requests." + "reason": "A BERT model addition is unrelated to an OpenAI weight-sparse transformer request." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT and LFM2.5 Audio are unrelated model families and modalities." + "reason": "NeoBERT (text) and LFM2.5 Audio are different model/support requests." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are separate model additions, not the same bug/change." + "reason": "Both are model-add requests, but for different concrete model families; not a duplicate change." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral TTS are different model-support requests." + "reason": "NeoBERT and Voxtral TTS are different models/modalities, so they should not be merged." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are unrelated model requests." + "reason": "DEIMv2 and Qwen3-TTS are unrelated requests with different tasks and implementation paths." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT models and Qwen3-ASR support are different model additions." + "reason": "ModernVBERT and Qwen3-ASR are different concrete model additions, not one bug fix." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT and Qwen3-TTS are distinct models and modalities." + "reason": "ModernVBERT models and Qwen3-TTS address different model families and modalities." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT and BERT-like models with RoPE overlap only loosely; not the same concrete request." + "reason": "ModernVBERT models vs a broad 'BERT-like models with RoPE' proposal is too loose; not the same concrete change." }, { "left": "issue:42503", "right": "issue:45027", "accept": false, - "reason": "ModernVBERT and Voxtral TTS are unrelated model-support requests." + "reason": "ModernVBERT and Voxtral TTS are unrelated model requests." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like models with RoPE and Voxtral TTS are different change requests." + "reason": "A general RoPE/BERT-like feature request is not the same as adding Voxtral TTS support." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Qwen3-TTS and Qwen3-ASR are sibling models, but separate concrete support requests." + "reason": "Qwen3-TTS and Qwen3-ASR are both speech-related but require different support paths and different concrete model integrations." }, { "left": "issue:43671", - "right": "issue:45027", + "right": "issue:43909", "accept": false, - "reason": "Qwen3-TTS and Voxtral TTS are different TTS models; not the same underlying change." + "reason": "Different audio models; same broad area, but not the same underlying change." }, { "left": "issue:43671", - "right": "issue:45177", + "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated modalities." + "reason": "Qwen3-TTS support is unrelated to the weight-sparse transformer proposal." }, { "left": "issue:43671", - "right": "issue:44121", + "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS and a weight-sparse transformer request are not the same issue." + "reason": "Both mention TTS, but they are different models and likely different integration work." }, { "left": "issue:43671", - "right": "issue:43909", + "right": "issue:45177", "accept": false, - "reason": "Qwen3-TTS and LFM2.5 Audio are different audio-model additions." + "reason": "TTS support and DeepSeek-OCR2 are different modalities and code paths." }, { "left": "issue:43837", - "right": "issue:44205", + "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and SAM3-LiteText are unrelated model requests." + "reason": "Both are audio-related, but Qwen3-ASR and LFM2.5 Audio are different concrete model requests." }, { "left": "issue:43837", - "right": "issue:45322", + "right": "issue:44205", "accept": false, - "reason": "Qwen3-ASR and EUPE are different model additions with different purposes." + "reason": "ASR support and SAM3-LiteText are unrelated model domains." }, { "left": "issue:43837", - "right": "issue:44121", - "accept": false, - "reason": "Qwen3-ASR and circuit-sparsity/circuitgpt are unrelated requests." - }, - { - "left": "issue:43909", - "right": "issue:44205", + "right": "issue:45322", "accept": false, - "reason": "LFM2.5 Audio and SAM3-LiteText are different modalities and model families." + "reason": "Qwen3-ASR and EUPE are different tasks/models with no shared underlying fix." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "Weight-sparse transformer/circuitgpt and EUPE are not the same underlying change." + "reason": "OpenAI weight-sparse transformer and EUPE are unrelated architecture/model requests." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral TTS are distinct audio-model support requests." + "reason": "Both are speech-related, but ASR vs TTS support for different models is not the same change." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "Qwen3-ASR and DeepSeek-OCR2 are unrelated models and tasks." + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model integrations." } ] }, @@ -14213,7 +14257,7 @@ "issue:43668|issue:44031", "issue:43668|issue:43813" ], - "prepared_review_unit_hash": "3bb24abe8d4df02a86b2f2eadca1bf6681310bace1506c6bd7a78cd470eec2f1", + "prepared_review_unit_hash": "f2801e03c578edf992d7bdb1778d3cb2b615c0c168d4268820def1ad304b6b77", "component_budget": { "node_count": 6, "item_count": 6, @@ -14235,60 +14279,60 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "No duplicates here: the cluster mixes unrelated issues across model architecture, config typing, loss plumbing, a typo, and tokenizer warnings.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue fits this cluster; the items describe different bugs in different components and code paths.", + "summary": "These issues are only superficially similar by wording; they cover different models, config annotations, loss plumbing, quantization typos, and a tokenizer warning regression, so none of the soft links look like true duplicates.", + "confidence": 0.78, + "canonical_issue_reason": "No true canonical issue: the set spans distinct bugs and code paths rather than one underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": "If one issue had to be kept as the most broadly impactful, issue #44031 stands out because it affects tokenizer warnings across versions, but it is still not a duplicate of the others.", + "best_issue_reason": "43240 is the best single anchor if one must be chosen: it is a concrete bug report with the most apparent cross-reference activity.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Both mention normalization, but one is a Llama4 architecture question and the other is a ModernBERT config type-hint bug; different models and different failure modes." + "reason": "Both mention normalization, but one is about Llama4TextL2Norm behavior and the other is a ModernBERT config type hint; different models and fixes." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Mask2Former ignore_value handling is unrelated to kwargs not being forwarded into a loss function; these are different code paths." + "reason": "Both concern training/loss plumbing, but Mask2Former ignore_value handling and kwargs forwarding to loss are different bugs in different paths." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "A model-specific ignore_value bug is not the same as a tokenizer regex warning regression; different subsystems entirely." + "reason": "Mask2Former ignore_value usage and tokenizer regex warnings are unrelated subsystems and symptoms." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "Ignore_value usage in Mask2Former is unrelated to the quantization_operation typo; one is behavioral, the other is a naming typo." + "reason": "Loss-argument handling in Mask2Former is unrelated to a quantization typo string." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2Former ignore_value behavior and ModernBERTConfig norm_eps typing are separate issues in different modules." + "reason": "A model loss-plumbing bug is not the same as an incorrect config type hint; different scope and remediation." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "Passing kwargs to loss calculation and correcting a config type hint are not the same bug or fix." + "reason": "Forwarding kwargs to a loss function and correcting a config annotation are separate issues with no shared code path." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "A config annotation error does not match a tokenizer regex-warning regression; no shared concrete code path." + "reason": "ModernBERTConfig type hints and tokenizer regex warnings are unrelated bugs." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "ModernBERT norm_eps typing and the orig_conversion.quantization_operation typo are unrelated fixes." + "reason": "A type-hint typo in config is unrelated to a quantization-operation typo." } ] }, @@ -14318,7 +14362,7 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "667360fea0bf6c210d61700a98016e949e15e9f4236c2cdae60064312e3b3faa", + "prepared_review_unit_hash": "cc88fe5f819a9686e46d27ffff34b0e5fd1258db9af892bba03b07313ca98249", "component_budget": { "node_count": 2, "item_count": 2, @@ -14340,18 +14384,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are unrelated: one is about block decoding, the other about adding DeepSeek v3.2 support. No duplicate relationship is evident.", - "confidence": 0.97, + "summary": "The two issues are not duplicates: one requests block decoding, the other requests DeepSeek v3.2 support. They touch different feature areas and describe different needs.", + "confidence": 0.95, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue #43037 is the clearer standalone request; #41084 is too terse and refers to a different topic.", + "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 shared code path or explicit linkage: block decoding vs model support." + "reason": "Different underlying requests: block decoding vs model/version support. No evidence they refer to the same bug or change." } ] }, @@ -14381,12 +14425,12 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "a4b318dc04141d89893eeabf7a3fae1fa009bd9182ff757c5abe2692922e8282", + "prepared_review_unit_hash": "fa3d23b0ceb21d8993c92f8de769fda26035581fd4a814ac8c3a8603df6abeb2", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, + "serialized_chars": 1583, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -14394,7 +14438,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, + "serialized_chars": 1583, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -14403,18 +14447,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are related to Trainer customizability, but they describe different problems: one is about `remove_unused_column` behavior, the other about FLOPS estimation for custom trainers.", - "confidence": 0.92, + "summary": "These two closed issues are about different Trainer concerns: one requests better customizability around `remove_unused_column`, while the other asks FLOPS estimation to account for custom trainers. They are related to Trainer internals but do not describe the same bug or change, so they should not be deduplicated.", + "confidence": 0.95, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43598 is the more focused standalone report because it names a specific Trainer setting and a concrete customization concern.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Related Trainer-area topics, but not the same bug: `remove_unused_column` customizability and FLOPS estimation for custom trainers are different code paths and fixes." + "reason": "Different underlying requests: customizability of `remove_unused_column` vs FLOPS estimation for custom trainers. Shared Trainer context is too broad to treat as a duplicate." } ] }, @@ -14444,7 +14488,7 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "05a7e3e6f391a42744c0870cba15fcf126b6cb74a0d251059c8b2c4a542d308d", + "prepared_review_unit_hash": "675303e5bfe517616d1b7b783857ea61ca0888674308221762b0eb745aa66918", "component_budget": { "node_count": 2, "item_count": 2, @@ -14466,18 +14510,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues request different model support in different API classes and for different Qwen variants, so they are not duplicates of the same underlying change.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "These two issues are about different Qwen-related feature gaps: one requests model registration for Qwen3-Omni on AutoModel/AutoModelForConditionalGeneration, while the other requests sequence-classification support for Qwen3.5. They are not duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical duplicate is evident; if one must be picked, issue #43684 is the broader API-registration request, but it is still a different feature from #44405.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue #43684 is the more general representative of a model-registration gap, but the cluster is not truly cohesive enough to treat either issue as a duplicate of the other.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different feature requests: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration vs Qwen3.5 sequence classification support. They target different model families and different class paths, so they should not be merged as the same bug/change." + "reason": "Different underlying requests: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration versus Qwen3.5 sequence-classification support. They affect different model/config paths and are not the same bug or change." } ] }, @@ -14514,76 +14558,76 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "f6709b66c523a904c3e27f229190c7c58c1b264d15b4bd4ed779f8b0726e1384", + "prepared_review_unit_hash": "d67c337eaeea102fe3702bda9c728c2b2ff119e2b77e8842a50bbbe9b7db1854", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4263, - "estimated_input_tokens": 1066, - "estimated_eval_tokens": 2388 + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 }, "budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4263, - "estimated_input_tokens": 1066, - "estimated_eval_tokens": 2388 + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The three \"huggingface \" issues look like duplicates of the same topic; \"Fett\" appears unrelated. No PRs are present.", - "confidence": 0.66, - "canonical_issue_reason": "Issue 43845 is the earliest of the near-identical \"huggingface \" reports and is the best representative of that duplicate set.", + "summary": "Three issues share the identical vague title \"huggingface\" and look duplicate-like; the \"Fett\" issue appears separate. I\u2019d cluster the three huggingface reports together, with 43845 as the canonical duplicate target and 43859 as the most informative representative.", + "confidence": 0.74, + "canonical_issue_reason": "43845 is the earliest of the repeated \"huggingface\" reports and cleanly represents the duplicate-like subgroup.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43845 is the cleanest canonical candidate: same title as the other likely duplicates, earliest created, and no evidence it differs materially.", + "best_issue_reason": "43859 has the most discussion/activity and likely the richest context, so it is the best triage-facing representative.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles and no concrete evidence of the same underlying issue; only generic soft similarity." + "reason": "Different titles and no concrete shared bug details; the weak similarity is not enough to treat them as the same issue." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles and no concrete evidence of the same underlying issue; only generic soft similarity." + "reason": "Same as above: \"Fett\" does not provide evidence of the same underlying problem as the huggingface reports." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no concrete evidence of the same underlying issue; only generic soft similarity." + "reason": "No substantive overlap beyond low soft similarity; insufficient evidence they describe the same bug." }, { "left": "issue:43845", "right": "issue:43846", "accept": true, - "reason": "Identical title, created minutes apart, and no contrary details; likely the same underlying issue report." + "reason": "Same title and very high similarity; they appear to be duplicate reports of the same issue." }, { "left": "issue:43845", "right": "issue:43859", "accept": true, - "reason": "Identical title and same apparent topic with no distinguishing body details; likely duplicate reports." + "reason": "Same title and strong similarity suggest the same underlying report, despite differing body length." }, { "left": "issue:43846", "right": "issue:43859", "accept": true, - "reason": "Identical title and same apparent topic with no distinguishing body details; likely duplicate reports." + "reason": "Same title and strong similarity; likely duplicate reports of the same issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The analyst stays within the packet: it only relies on identical titles, close timestamps for 43845/43846, and the lack of distinguishing evidence. The rejection of 43835 as unrelated is conservative, and the summary does not overclaim beyond what\u2019s visible. No PRs are present, as stated." + "feedback": "The clustering is grounded in the packet and stays appropriately cautious: 43845/43846/43859 share the identical title and the strongest soft-similarity scores, while 43835 has a different title and only weak similarity. The chosen canonical/representative issue assignments are conservative enough." }, "evaluator_used": true, "retried": false, @@ -14621,7 +14665,7 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "72526afb4207a6eb068f7dcabbd7bc70e8d6d5a6c815e8cdd6982467e30dc6b7", + "prepared_review_unit_hash": "0f1beca16484909016a6b8dfd2eaec788144cc722c5da7d7ff132f5e30fcde53", "component_budget": { "node_count": 5, "item_count": 5, @@ -14643,66 +14687,66 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are mostly spam or generic placeholder issues, with one vague documentation suggestion. None describe the same concrete bug or change, so no duplicates should be merged.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "These are mostly unrelated low-information/spam issues, with one generic documentation request. There\u2019s no solid evidence of a shared underlying bug or change across the set, so this cluster should not be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 44418 is the only non-spam-sounding item, so it is the least-bad anchor if a canonical issue must be chosen, but it still does not appear related to the others.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44418 is the only non-spam, somewhat descriptive item, but it is still too generic to serve as a strong canonical issue.", + "best_issue_reason": "Issue 44418 is the most substantive of the group and the only one with a real topic in the title; the rest are spam-like placeholders.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are spam-like placeholders; no shared underlying bug or change is described." + "reason": "Both are spam-like placeholders; no concrete shared problem is described." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Both are spam-like placeholders; no shared underlying bug or change is described." + "reason": "Both are spam-like placeholders; no evidence they describe the same issue." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "One is spam, the other is a generic documentation suggestion; no concrete match." + "reason": "One is spam, the other is a generic documentation suggestion; different intent and content." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder vs a different generic quoted phrase; no same issue is evident." + "reason": "Spam placeholder versus a specific quoted-name topic; not the same underlying issue." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "Generic documentation request and spam placeholder are not the same underlying issue." + "reason": "Generic documentation request and spam placeholder are unrelated." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "Generic documentation request and spam placeholder are not the same underlying issue." + "reason": "Generic documentation request and spam placeholder are unrelated." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Both are spam placeholders, but there is no evidence they refer to the same concrete problem." + "reason": "Both are spam-like placeholders, but there is no concrete bug/change to unify." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder vs unrelated quoted text; no shared bug or change." + "reason": "Spam placeholder versus a specific quoted-name topic; no shared underlying report." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder vs unrelated quoted text; no shared bug or change." + "reason": "Spam placeholder versus a specific quoted-name topic; no shared underlying report." } ] }, @@ -14732,7 +14776,7 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "a7c7198dcd4a6f02caa3a5cff316d765bf63540e07aeb4bb6d09f7fd252f5563", + "prepared_review_unit_hash": "f7bfb66cdd8102dca285d7dedd8a82997bac8fd3aa2e83b044bcf1ab11d2b5c9", "component_budget": { "node_count": 2, "item_count": 2, @@ -14754,24 +14798,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are likely duplicates: they share the exact same title, have similar-sized bodies, and there are no differentiating links, targets, or filenames. #45145 is the best canonical issue because it was created first and appears to be the earlier instance of the same report.", - "confidence": 0.91, - "canonical_issue_reason": "#45145 and #45151 have identical titles and no differentiating metadata; #45145 is earlier and is the best canonical issue.", + "summary": "These two closed issues are strong near-duplicates based on identical titles and very similar bodies, but the packet does not prove they are the same underlying report. #45145 is the better canonical representative because it was created first and has an inbound reference.", + "confidence": 0.78, + "canonical_issue_reason": "#45145 is earlier and already has an inbound reference, so it is the best canonical issue to retain.", "canonical_pr_reason": null, - "best_issue_reason": "#45145 is the best issue candidate because it is the earliest-created likely duplicate report.", + "best_issue_reason": "#45145 is the strongest issue to keep as canonical because it was created first and has more linkage than #45151.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", "accept": false, - "reason": "The pair looks like a likely duplicate based on identical titles and soft similarity, but the packet only provides a moderate Jaccard signal and does not establish that the bodies are the same report." + "reason": "Identical titles and soft-similarity make this a likely duplicate, but the evidence does not establish identical body content or confirm the same underlying issue. Conservative triage should not mark it as an accepted duplicate on this packet alone." } ] }, "evaluator_result": { "accept": false, - "feedback": "The duplicate/canonical call is mostly grounded by the identical titles, similar body sizes, and lack of differentiating links/targets, but the soft-edge verdict is a bit too strong. The packet only provides a soft-similarity signal (jaccard ~0.63) and does not establish that the bodies are the same or that they are definitely duplicates. Please soften the wording to \u201clikely duplicate\u201d rather than \u201ceffectively the same report content,\u201d and avoid an unconditional accept on the edge." + "feedback": "The canonical-choice rationale is partly grounded (45145 is earlier and has one inbound reference), but the soft-edge acceptance is too strong for the evidence. The packet only shows identical titles and a soft-similarity signal; it does not establish that the bodies are the same or that these are definitely duplicates. Also, \"slightly more linkage/activity\" is not supported as stated because 45151 has one discussion activity while 45145 does not. A more conservative verdict would be to mark the pair as a likely duplicate rather than an accepted duplicate." }, "evaluator_used": true, "retried": true, @@ -14836,7 +14880,7 @@ { "label": "pull_request", "component_index": 1, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -14844,7 +14888,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-39895-12", + "cluster_id": "cluster-39895-11", "nodes": [ "pull_request:39895", "pull_request:42668", @@ -14856,8 +14900,7 @@ "pull_request:44827", "pull_request:45170", "pull_request:45213", - "pull_request:45435", - "pull_request:45454" + "pull_request:45435" ], "soft_pairs": [ "pull_request:39895|pull_request:43424", @@ -14867,177 +14910,170 @@ "pull_request:43424|pull_request:44827", "pull_request:43424|pull_request:45435", "pull_request:43424|pull_request:43823", - "pull_request:43823|pull_request:45454", "pull_request:43823|pull_request:45435", "pull_request:43823|pull_request:43995", + "pull_request:43823|pull_request:44827", "pull_request:43995|pull_request:45435", "pull_request:43995|pull_request:45213", "pull_request:43995|pull_request:44827", "pull_request:44827|pull_request:45170", - "pull_request:43823|pull_request:44827", "pull_request:44827|pull_request:45213", "pull_request:45213|pull_request:45435", "pull_request:43823|pull_request:45213", "pull_request:44827|pull_request:45435", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "863d9589a643605f8b437ae00771fe54ea8dec929c1c6b16248446a94c7d71e8", + "prepared_review_unit_hash": "39ec3bb4ea751e1f7b383e73ec3002f0a5c4b0afaaad2bba16c7a50c33869f3d", "component_budget": { - "node_count": 12, - "item_count": 12, - "soft_pair_count": 20, - "serialized_chars": 34003, - "estimated_input_tokens": 8501, - "estimated_eval_tokens": 17258 + "node_count": 11, + "item_count": 11, + "soft_pair_count": 19, + "serialized_chars": 31458, + "estimated_input_tokens": 7865, + "estimated_eval_tokens": 15986 }, "budget": { - "node_count": 12, - "item_count": 12, - "soft_pair_count": 20, - "serialized_chars": 34003, - "estimated_input_tokens": 8501, - "estimated_eval_tokens": 17258 + "node_count": 11, + "item_count": 11, + "soft_pair_count": 19, + "serialized_chars": 31458, + "estimated_input_tokens": 7865, + "estimated_eval_tokens": 15986 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Most items in the cluster are unrelated PRs spanning new models, test coverage, refactors, and docs. The only confirmed duplicate-style overlap is the `fixed_cross_entropy` pair in `loss_utils.py` (43251/43254). 43251 is a reasonable representative, but 43254 is also a direct tracked fix, so the packet does not support exclusivity language.", - "confidence": 0.95, + "summary": "This set is mostly unrelated PRs spanning model additions, refactors, docs, and tests. The only true duplicate-like pair is the two `fixed_cross_entropy` fixes for issue #43240, which target the same function and underlying missing kwargs support.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43251 is a good representative of the only confirmed duplicate pair: it updates `fixed_cross_entropy` in `loss_utils.py` to forward supported `cross_entropy` kwargs, matching the same underlying fix as 43254.", + "canonical_pr_reason": "PR #43251 is the clearest canonical representative: it explicitly targets issue #43240 and fixes `fixed_cross_entropy` to forward missing `cross_entropy` kwargs in the same code path as #43254.", "best_issue_reason": null, - "best_pr_reason": "PR 43251 is a suitable global representative because it sits in the only confirmed duplicate pair and clearly shows the loss-utils fix, though 43254 is also a strong match and not meaningfully less representative from the packet alone.", + "best_pr_reason": "PR #43251 is the best representative overall because it is tightly scoped, issue-linked, and addresses the same concrete loss-function bug as #43254.", "soft_edge_verdicts": [ { "left": "pull_request:39895", "right": "pull_request:43424", "accept": false, - "reason": "Different changes: new VideoPrism model integration vs an ExecuTorch dynamic-shape export test." + "reason": "Unrelated changes: new VideoPrism model/docs versus an ExecuTorch dynamic-shape test." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Processor auto-loading robustness changes are unrelated to adding MobileLLM model code." + "reason": "Different workstreams: processor auto-loading robustness versus adding MobileLLM." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "Loss utility kwargs handling is unrelated to adding a new model implementation." + "reason": "Loss helper kwargs fix is unrelated to adding MobileLLM." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "`fixed_cross_entropy` kwargs support and Mistral4 test fixes are different code paths and different bugs." + "reason": "Same broad library area, but one is a loss-function fix and the other is Mistral4 test/model work." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "ExecuTorch dynamic-shape export tests are unrelated to Mistral4 model/test changes." + "reason": "ExecuTorch export test and Mistral4 tests are different concrete changes." }, { "left": "pull_request:43424", "right": "pull_request:45435", "accept": false, - "reason": "ExecuTorch export coverage and Whisper tokenization bounds handling are separate problems." + "reason": "Dynamic-shape export test versus Whisper tokenizer bugfix; no shared underlying change." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "Export test coverage is unrelated to a new model addition." + "reason": "Different features: ExecuTorch testing versus new MobileLLM model support." }, { "left": "pull_request:43823", - "right": "pull_request:45454", + "right": "pull_request:45435", "accept": false, - "reason": "Adding MobileLLM and Gemma4 training-mask logic are unrelated changes." + "reason": "MobileLLM model addition and Whisper tokenizer fix are unrelated." }, { "left": "pull_request:43823", - "right": "pull_request:45435", + "right": "pull_request:43995", "accept": false, - "reason": "MobileLLM model addition and Whisper tokenizer indexing fixes are different artifacts with no shared bug." + "reason": "MobileLLM addition versus Falcon output-interface refactor; not the same bug/change." }, { "left": "pull_request:43823", - "right": "pull_request:43995", + "right": "pull_request:44827", "accept": false, - "reason": "New model addition vs Falcon output-collection refactor; different scope and issue." + "reason": "New MobileLLM model work is unrelated to Mistral4 test fixes." }, { "left": "pull_request:43995", "right": "pull_request:45435", "accept": false, - "reason": "Falcon output-collection refactor is not the same as Whisper tokenization bounds handling." + "reason": "Falcon refactor and Whisper tokenizer fix touch different code paths and problems." }, { "left": "pull_request:43995", "right": "pull_request:45213", "accept": false, - "reason": "Falcon refactor and model-creation skill content are unrelated." + "reason": "Falcon model refactor is unrelated to the model-creation skill draft PR." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Falcon standardized output collection and Mistral4 test fixes do not address the same code-path problem." + "reason": "Falcon standardized output collection refactor and Mistral4 tests are not the same underlying change." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Mistral4 test fixes and CLIP-like `layrnorm` renaming are different changes with no shared bug." - }, - { - "left": "pull_request:43823", - "right": "pull_request:44827", - "accept": false, - "reason": "Adding MobileLLM is unrelated to Mistral4 tests." + "reason": "Mistral4 test/model fixes versus a broad `layrnorm` rename across CLIP-like models; different bugs." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "Mistral4 test fixes and the model-creation skill PR are not the same underlying change." + "reason": "Mistral4 test fixes and the model-creation skill PR are unrelated." }, { "left": "pull_request:45213", "right": "pull_request:45435", "accept": false, - "reason": "A meta skill/documentation PR is unrelated to Whisper tokenizer bug fixing." + "reason": "A meta skill/documentation PR is not the same as a Whisper tokenizer bugfix." }, { "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "MobileLLM model addition and model-creation skill content are unrelated." + "reason": "MobileLLM model addition and the model-creation skill draft do not fix the same concrete code-path problem." }, { "left": "pull_request:44827", "right": "pull_request:45435", "accept": false, - "reason": "Mistral4 test fixes and Whisper tokenizer indexing fix are separate code paths." + "reason": "Mistral4 test fixes and Whisper tokenizer bugfix are unrelated." }, { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Both PRs fix `fixed_cross_entropy` in `loss_utils.py` to accept and pass through supported `cross_entropy` kwargs; they target the same underlying bug and could plausibly be merged into one PR." + "reason": "Both PRs fix the same `fixed_cross_entropy` kwargs forwarding bug in `loss_utils.py` for issue #43240, and could plausibly be merged into one change." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The pairwise verdicts are mostly conservative and grounded, but the summary/representative reasons overstate the evidence: PR 43254 also has an explicit issue link and edits the same `loss_utils.py` `fixed_cross_entropy` path, so 43251 is not the only item with a clear tracked bug fix. Keep the duplicate-pair claim, but avoid exclusivity language that the packet does not support." + "accept": true, + "feedback": "The evaluation is well-grounded in the packet. The summary stays conservative, and the only accepted soft-edge pair, PR #43251 and PR #43254, is supported by the shared `loss_utils.py` change and the same issue target (#43240). The other pairwise rejections are also appropriately cautious and do not overclaim mergeability." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -15045,7 +15081,7 @@ { "label": "pull_request", "component_index": 2, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15061,7 +15097,7 @@ "soft_pairs": [ "pull_request:40546|pull_request:42781" ], - "prepared_review_unit_hash": "e196e240796cd841be3da2a32d71752dd9623b4476286174e266e5a594f96ac8", + "prepared_review_unit_hash": "dc199947d14aa90df5aa9dda5c5859c2b9fa98c59bca603d9e9847506b0bf208", "component_budget": { "node_count": 2, "item_count": 2, @@ -15083,18 +15119,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests are about VibeVoice, but they target different scopes: 40546 adds the base VibeVoice model integration, while 42781 adds a separate VibeVoice Realtime extension plus the acoustic tokenizer. They overlap in shared model/auto-registration files, but they are not the same concrete change.", - "confidence": 0.92, + "summary": "These are related VibeVoice PRs, but they are not duplicates: #40546 adds the initial VibeVoice implementation, while #42781 adds a broader realtime variant plus an acoustic tokenizer and additional generation/model changes. Overlap is mostly shared model scaffolding and docs, not the same concrete fix/change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 40546 is the better canonical representative because it is the original, non-draft base VibeVoice implementation with broader review activity, while 42781 is a later draft focused on the realtime extension and tokenizer add-on.", + "canonical_pr_reason": "#42781 is the broader and later VibeVoice PR, covering the realtime path and acoustic tokenizer in addition to the base model, so it is the better representative if one PR must be chosen for the cluster.", "best_issue_reason": null, - "best_pr_reason": "40546 is the strongest standalone PR for the cluster: it covers the core VibeVoice integration and documentation, and it is more complete/stable than the draft realtime follow-up.", + "best_pr_reason": "#42781 is the most complete VibeVoice PR in this set and subsumes more of the related implementation surface, making it the better anchor candidate for triage.", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "Different feature sets: 40546 implements the base VibeVoice model, while 42781 adds VibeVoice Realtime and an acoustic tokenizer. Shared files are expected from related integration work, but this is not the same underlying change and would not plausibly be merged as one PR." + "reason": "Reject: they share VibeVoice infrastructure files, but the PRs target different changesets\u2014initial VibeVoice support vs. realtime + acoustic tokenizer expansion. They do not look like the same underlying bug or a single mergeable fix." } ] }, @@ -15108,7 +15144,7 @@ { "label": "pull_request", "component_index": 3, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15124,12 +15160,12 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "d8252ea3c274e117ff946f6c34b39caced12b39054f28d617718260802386221", + "prepared_review_unit_hash": "fea9054a212e574d7670f8c791e82c42a65e33d59924b7cbcf1105f30cfd44c2", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4676, + "serialized_chars": 4675, "estimated_input_tokens": 1169, "estimated_eval_tokens": 2594 }, @@ -15137,7 +15173,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4676, + "serialized_chars": 4675, "estimated_input_tokens": 1169, "estimated_eval_tokens": 2594 }, @@ -15146,18 +15182,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are unrelated: one adds TimesFM covariate forecasting, the other adds DINOv3 image-classification support. They touch different models, different code paths, and solve different problems, so they should not be clustered as duplicates.", + "summary": "The cluster contains two unrelated pull requests: one adds TimesFM covariate-based forecasting, the other adds DINOv3 image classification support. They do not fix the same bug or implement the same change, so they should not be merged as duplicates.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No single canonical PR fits this cluster: the two PRs target different model families and different features, with no shared underlying code-path problem.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "PR #40755 is the broader, more feature-complete TimesFM change, but it is not a representative duplicate of the DINOv3 classification PR; overall, neither PR is a good cluster-wide canonical.", "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different model families and features: TimesFM covariate regression vs. DINOv3 image classification. The overlap is only broad ML/modeling/docs work, not the same underlying change or bug." + "reason": "Different models and features: TimesFM covariate forecasting vs. DINOv3 image classification support. They are not the same underlying change and would not be merged into one PR." } ] }, @@ -15171,7 +15207,7 @@ { "label": "pull_request", "component_index": 4, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15187,54 +15223,57 @@ "soft_pairs": [ "pull_request:41356|pull_request:44339" ], - "prepared_review_unit_hash": "f5132dd14a644819948ba96a881537ef7205739795a383967c1ad5cebe2c9b4c", + "prepared_review_unit_hash": "a3c962d2540926578e2da616c4d35ea269f16dd241073ecbf1aac9c482085f67", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5929, - "estimated_input_tokens": 1483, - "estimated_eval_tokens": 3222 + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5929, - "estimated_input_tokens": 1483, - "estimated_eval_tokens": 3222 + "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": "Both pull requests are about adding DEIMv2 to Transformers, but they are not duplicates: the first is a smaller initial model/image-processor drop, while the second is a broader follow-up with losses, auto-mappings, conversion tooling, and modular code.", - "confidence": 0.92, + "summary": "Both pull requests are about adding DEIMv2 to Transformers. PR 44339 is the more complete, later integration and appears to supersede the earlier basic implementation in 41356.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #44339 is the more complete DEIMv2 implementation and integration: it adds core model code plus loss, auto mappings, converter, modular support, and config validation, making it the better canonical representative of the feature.", + "canonical_pr_reason": "44339 is the more complete DEIMv2 integration and supersets 41356 with broader model wiring, auto-mapping updates, conversion tooling, loss support, and config checks.", "best_issue_reason": null, - "best_pr_reason": "PR #44339 is the best overall PR because it covers the broadest and most integrated DEIMv2 support and is closer to a shippable Transformers addition than the smaller initial PR.", + "best_pr_reason": "44339 is the strongest representative of the cluster because it covers the full DEIMv2 addition rather than only the initial model/image-processor scaffold.", "soft_edge_verdicts": [ { "left": "pull_request:41356", "right": "pull_request:44339", - "accept": false, - "reason": "Not a duplicate PR: #41356 is a smaller model/image-processor/test addition, while #44339 expands the feature with loss, auto-mapping, conversion, and modular integration. They overlap on the model but not the same concrete code-path change." + "accept": true, + "reason": "Both PRs implement the same underlying change: adding DEIMv2 support. 44339 is effectively a superset of 41356, so they are close enough to treat as the same feature change." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs target issue 41211 and add DEIMv2, with 44339 clearly extending the earlier scaffold in 41356 via auto-mapping, conversion, loss, and config-related changes. The soft-edge verdict is conservative enough because the two changes look like mergeable iterations of the same feature work." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, { "label": "pull_request", "component_index": 5, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15250,12 +15289,12 @@ "soft_pairs": [ "pull_request:41797|pull_request:45186" ], - "prepared_review_unit_hash": "b7c161bac82eac2e92a0510c22272f289dd1beb0bef01c32f77a15f96b1194b1", + "prepared_review_unit_hash": "7443f734b60321912bc7423de9cf15dee6ff9d8a12b1d8eeed23ba618e973839", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6723, + "serialized_chars": 6721, "estimated_input_tokens": 1681, "estimated_eval_tokens": 3618 }, @@ -15263,7 +15302,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6723, + "serialized_chars": 6721, "estimated_input_tokens": 1681, "estimated_eval_tokens": 3618 }, @@ -15272,18 +15311,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate model-addition PRs: one for DeepSeek OCR and one for Isaac. They share boilerplate model-integration files, but they do not target the same code path or change set, so they should not be deduplicated.", - "confidence": 0.97, + "summary": "These are two separate model-introduction PRs (DeepSeek OCR vs Isaac). They share the same Transformers integration pattern and some registry/doc files, but they target different models and different code paths, so they are not duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the PRs add different models with different configs, processors, tests, and docs, so they are not duplicates.", + "canonical_pr_reason": "No canonical PR: the PRs implement different models and cannot be collapsed into one duplicate change.", "best_issue_reason": null, - "best_pr_reason": "PR 41797 is the more established representative of the cluster because it has much higher review activity and appears more mature, but it is not a duplicate of 45186.", + "best_pr_reason": "No single best PR for deduping here; neither PR subsumes the other because they add unrelated model integrations.", "soft_edge_verdicts": [ { "left": "pull_request:41797", "right": "pull_request:45186", "accept": false, - "reason": "Reject: both are model additions, but for different models (DeepSeek OCR vs Isaac). Shared auto-registration and docs/test scaffolding are expected boilerplate, not evidence of the same underlying change." + "reason": "Reject: both are new-model integration PRs, but for different models (DeepSeek OCR vs Isaac) with distinct configs, models, processors, tests, and docs. Shared registry/docs scaffolding does not indicate the same underlying bug or change." } ] }, @@ -15297,7 +15336,7 @@ { "label": "pull_request", "component_index": 6, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15313,12 +15352,12 @@ "soft_pairs": [ "pull_request:42166|pull_request:45426" ], - "prepared_review_unit_hash": "acf5732467185d203c066cded3763a66b5f5ee0ea210f605ae2ab497a94de842", + "prepared_review_unit_hash": "33b1ba550b4994f8e2a9729e5c43b0b60e3a40ae4e20035663a4a4352064cb3a", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5407, + "serialized_chars": 5408, "estimated_input_tokens": 1352, "estimated_eval_tokens": 2960 }, @@ -15326,7 +15365,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5407, + "serialized_chars": 5408, "estimated_input_tokens": 1352, "estimated_eval_tokens": 2960 }, @@ -15335,18 +15374,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated model-addition PRs: one adds InternVLFlash, the other adds AXK1. Their overlap is limited to generic Transformers registration plumbing, so they should not be clustered as duplicates.", - "confidence": 0.96, + "summary": "The two pull requests are unrelated model-addition PRs: one adds InternVLFlash, the other adds AXK1. They share only generic auto-registration plumbing and are not the same underlying change.", + "confidence": 0.97, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "No strong canonical PR exists here; if one must be chosen as the cluster representative, PR #42166 is slightly more substantial and established (more files changed and higher review activity), but it is not a duplicate of PR #45426.", "soft_edge_verdicts": [ { "left": "pull_request:42166", "right": "pull_request:45426", "accept": false, - "reason": "Different underlying features/models (InternVLFlash vs AXK1). Shared files are only broad auto-registration and package wiring, which is not enough to indicate the same bug or change." + "reason": "Reject: these are different model-feature PRs for different architectures (InternVLFlash vs AXK1). Shared auto-registration files are incidental and do not indicate the same concrete code-path fix or change." } ] }, @@ -15360,7 +15399,7 @@ { "label": "pull_request", "component_index": 7, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15368,12 +15407,13 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-42227-4", + "cluster_id": "cluster-42227-5", "nodes": [ "pull_request:42227", "pull_request:43291", "pull_request:43665", - "pull_request:44815" + "pull_request:44815", + "pull_request:45574" ], "soft_pairs": [ "pull_request:42227|pull_request:43291", @@ -15381,72 +15421,86 @@ "pull_request:42227|pull_request:44815", "pull_request:43291|pull_request:43665", "pull_request:43291|pull_request:44815", - "pull_request:43665|pull_request:44815" + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:45574", + "pull_request:44815|pull_request:45574" ], - "prepared_review_unit_hash": "f25fee4e50631439d99feadf245a793345b47ad1d67a1518a798215c146d0887", + "prepared_review_unit_hash": "e4d67011c963b429746bb6e144ec1c0e6cf89f2279da7af0cbff645a5afddd40", "component_budget": { - "node_count": 4, - "item_count": 4, - "soft_pair_count": 6, - "serialized_chars": 10168, - "estimated_input_tokens": 2542, - "estimated_eval_tokens": 5340 + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 13477, + "estimated_input_tokens": 3370, + "estimated_eval_tokens": 6996 }, "budget": { - "node_count": 4, - "item_count": 4, - "soft_pair_count": 6, - "serialized_chars": 10168, - "estimated_input_tokens": 2542, - "estimated_eval_tokens": 5340 + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 13477, + "estimated_input_tokens": 3370, + "estimated_eval_tokens": 6996 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are only superficially similar. They span unrelated areas: Whisper ASR forwarding, Whisper tokenizer/test cleanup, ad hoc test-file edits, and FP8 dequant/model-loading changes. None of the pairs look like the same underlying bug or change.", - "confidence": 0.96, + "summary": "All candidates are pull requests; none are strong duplicate/merge candidates with each other. The Whisper PR and the Whisper tests/tokenization PR touch related area but fix different behaviors, while the remaining PRs are clearly unrelated or broad cosmetic/test-only changes.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR #42227 is the clearest focused fix: it directly targets an explicit issue, changes one concrete ASR code path, and has a narrowly scoped diff for Whisper return_language handling.", "best_issue_reason": null, - "best_pr_reason": "PR 42227 is the strongest standalone candidate: it fixes a concrete Whisper pipeline bug, has an explicit issue target, and the diff is coherent and narrowly scoped.", + "best_pr_reason": "PR #42227 is the best standalone candidate because it has a specific bug report link, a small targeted change, and a clear functional purpose compared with the broader or test-only alternatives.", "soft_edge_verdicts": [ { "left": "pull_request:42227", "right": "pull_request:43291", "accept": false, - "reason": "Both mention Whisper, but 42227 fixes `return_language` propagation in the ASR pipeline while 43291 changes tokenizer decode behavior and tests. Different code paths and different bugs." + "reason": "Both are Whisper-related, but #42227 fixes ASR pipeline return_language propagation while #43291 changes tokenizer decode/test behavior; different code paths and different bugs." }, { "left": "pull_request:42227", "right": "pull_request:43665", "accept": false, - "reason": "42227 is a real Whisper pipeline fix; 43665 is just test-file edits with unrelated assertions. Not the same change or bug." + "reason": "#42227 is a Whisper runtime fix; #43665 is a broad test-only edit across CLIP/ViT with no shared bug or concrete code-path overlap." }, { "left": "pull_request:42227", "right": "pull_request:44815", "accept": false, - "reason": "Whisper ASR pipeline bug vs FP8 dequant/model-loading work. Completely different subsystems and problems." + "reason": "#42227 addresses Whisper return_language in the pipeline; #44815 is a larger FP8/dequant loading change. They are unrelated fixes." }, { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "43291 is Whisper tokenizer logic; 43665 is unrelated test churn. No shared underlying bug." + "reason": "#43291 is a Whisper tokenizer behavior change; #43665 only alters tests in unrelated vision models." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Whisper decode changes and Mistral/FP8 dequant changes are unrelated." + "reason": "#43291 concerns Whisper decode logic, while #44815 concerns FP8 dequant/loading infrastructure. Different subsystems and bugs." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "Test-only modifications do not match the FP8 dequant/model-loading fix; no common code-path issue." + "reason": "#43665 is test noise in CLIP/ViT; #44815 is a substantive quantization/loading fix. No shared underlying change." + }, + { + "left": "pull_request:43665", + "right": "pull_request:45574", + "accept": false, + "reason": "#43665 is unrelated vision test edits; #45574 adds a new model/docs/registry support. Different changes entirely." + }, + { + "left": "pull_request:44815", + "right": "pull_request:45574", + "accept": false, + "reason": "#44815 is an FP8/dequant fix; #45574 is new model onboarding plus docs and registry updates. Not the same bug or change." } ] }, @@ -15460,7 +15514,7 @@ { "label": "pull_request", "component_index": 8, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15476,12 +15530,12 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "bc0e94a1250355b6afb609c28a97d1bf88a683f0d16dd1092d18b189d4345981", + "prepared_review_unit_hash": "84970cd01fbd704304ff489ebaab2dd47fc3cda0aba17a576072b0f522501473", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5935, + "serialized_chars": 5934, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -15489,7 +15543,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5935, + "serialized_chars": 5934, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -15498,18 +15552,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs are broad workflow dependency bumps, but they upgrade different action families and fix different compatibility concerns, so they are not the same change.", - "confidence": 0.89, + "summary": "Two open PRs both update GitHub Actions dependencies, but they target different action families and different maintenance goals. They are not the same underlying change and should not be merged as duplicates.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the broader, more central upgrade focused on GitHub Actions Node 24 compatibility across many workflows, making it the better representative of this cluster.", + "canonical_pr_reason": "PR 42975 is the broader, more central maintenance change: it updates core GitHub Actions used across many workflows for Node 24 compatibility.", "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the best representative because its scope is broader and more clearly describes the underlying maintenance effort; PR 42976 is a narrower Docker action version bump.", + "best_pr_reason": "PR 42975 is the better representative PR for the cluster because it has the wider impact and clearer compatibility-driven scope.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "They touch overlapping workflow files, but 42975 upgrades core GitHub Actions like checkout/cache/upload-artifact for Node 24 compatibility, while 42976 bumps docker/build-push-action and a few other action versions. These are related maintenance chores, not the same concrete bug or change." + "reason": "Both are workflow dependency upgrades, but 42975 updates actions/checkout, actions/cache, and upload-artifact for Node 24 compatibility, while 42976 updates docker/build-push-action and related workflow versions. They touch overlapping CI files but fix different action-version maintenance issues, not the same concrete bug or change." } ] }, @@ -15523,7 +15577,7 @@ { "label": "pull_request", "component_index": 9, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15531,18 +15585,12 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-42978-10", + "cluster_id": "cluster-42978-4", "nodes": [ "pull_request:42978", "pull_request:43448", "pull_request:43451", - "pull_request:43838", - "pull_request:43973", - "pull_request:44601", - "pull_request:45157", - "pull_request:45283", - "pull_request:45490", - "pull_request:45532" + "pull_request:45490" ], "soft_pairs": [ "pull_request:42978|pull_request:43448", @@ -15550,128 +15598,72 @@ "pull_request:42978|pull_request:45490", "pull_request:43448|pull_request:43451", "pull_request:43448|pull_request:45490", - "pull_request:43448|pull_request:45532", - "pull_request:43451|pull_request:45490", - "pull_request:43451|pull_request:45532", - "pull_request:43838|pull_request:45283", - "pull_request:43973|pull_request:45283", - "pull_request:44601|pull_request:45532", - "pull_request:45157|pull_request:45532", - "pull_request:45283|pull_request:45532", - "pull_request:45490|pull_request:45532" + "pull_request:43451|pull_request:45490" ], - "prepared_review_unit_hash": "0a88a5201bb313bc4c0be06994f5178f6d6dade169f41db1af0b18b339fad3dd", + "prepared_review_unit_hash": "0ae72507573ba71181825e6c0fc141d452861230c5bdf97f4275d46b8d8a1c13", "component_budget": { - "node_count": 10, - "item_count": 10, - "soft_pair_count": 14, - "serialized_chars": 32327, - "estimated_input_tokens": 8082, - "estimated_eval_tokens": 16420 + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14136, + "estimated_input_tokens": 3534, + "estimated_eval_tokens": 7324 }, "budget": { - "node_count": 10, - "item_count": 10, - "soft_pair_count": 14, - "serialized_chars": 32327, - "estimated_input_tokens": 8082, - "estimated_eval_tokens": 16420 + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14136, + "estimated_input_tokens": 3534, + "estimated_eval_tokens": 7324 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is a set of unrelated Transformers PRs that happen to share boilerplate model-registration, docs, and auto-mapping changes. They are not duplicates of the same bug or feature, so no mergeable canonical group emerges.", - "confidence": 0.98, + "summary": "All four items are separate model-addition PRs that only overlap on Transformers boilerplate (docs, auto mappings, tests). They do not describe the same underlying change, so no soft pair should be merged.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: the items span different model additions and unrelated infrastructure changes, with only shared framework boilerplate in common.", + "canonical_pr_reason": "No single canonical PR stands out as a duplicate target; if forced to pick one representative, 43451 is the most substantial/complete, but it is still a distinct Molmo2 addition rather than a duplicate of the others.", "best_issue_reason": null, - "best_pr_reason": "No PR in this cluster is a strong global canonical choice because the overlap is superficial; the closest items still target different models/code paths.", + "best_pr_reason": "43451 is the strongest standalone PR in the set by scope and review activity, but it is not duplicate-equivalent to the others.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different model additions (ViT NEPA vs Molmo); overlap is just docs/auto-registration boilerplate." + "reason": "ViT NEPA vs Molmo: different model families and different implementation files, so this is not the same code change." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "Different model additions (ViT NEPA vs Molmo2); shared files are generic registration/docs, not the same change." + "reason": "ViT NEPA vs Molmo2: unrelated model additions; shared auto/docs scaffolding is too generic to indicate a duplicate." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "Different models (ViT NEPA vs CTSM); only shared Transformers scaffold changes." + "reason": "ViT NEPA vs CTSM are distinct model integrations with different configs/modeling paths; not the same underlying feature." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are distinct model support PRs, not one underlying fix." + "reason": "Molmo vs Molmo2 are related by naming but they add different models and codepaths, not one mergeable fix/change." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo and CTSM are unrelated model additions; shared auto/docs edits are incidental." - }, - { - "left": "pull_request:43448", - "right": "pull_request:45532", - "accept": false, - "reason": "Molmo and SLANet target different models and code paths; only boilerplate overlap." + "reason": "Molmo and CTSM are unrelated model additions; overlap is limited to shared registration/docs updates." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 vs CTSM are unrelated model-support changes." - }, - { - "left": "pull_request:43451", - "right": "pull_request:45532", - "accept": false, - "reason": "Molmo2 and SLANet are separate model additions; no shared concrete bug/change." - }, - { - "left": "pull_request:43838", - "right": "pull_request:45283", - "accept": false, - "reason": "Qwen3-ASR support and Qwen3.5 GGUF loading are different features/code paths." - }, - { - "left": "pull_request:43973", - "right": "pull_request:45283", - "accept": false, - "reason": "Lfm2Audio model support is unrelated to Qwen3.5 GGUF loading support." - }, - { - "left": "pull_request:44601", - "right": "pull_request:45532", - "accept": false, - "reason": "Distributed pipeline-parallel support is infrastructure work, not the SLANet model addition." - }, - { - "left": "pull_request:45157", - "right": "pull_request:45532", - "accept": false, - "reason": "PrismML Bonsai GGUF handling and SLANet model support are unrelated changes." - }, - { - "left": "pull_request:45283", - "right": "pull_request:45532", - "accept": false, - "reason": "Qwen3.5 GGUF loading support and SLANet model support do not fix the same code-path problem." - }, - { - "left": "pull_request:45490", - "right": "pull_request:45532", - "accept": false, - "reason": "CTSM and SLANet are different model support PRs; similarity is only in shared model-template files." + "reason": "Molmo2 vs CTSM are different model implementations with no concrete shared bug or change to merge together." } ] }, @@ -15685,7 +15677,7 @@ { "label": "pull_request", "component_index": 10, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15704,52 +15696,52 @@ "pull_request:43488|pull_request:45476", "pull_request:43532|pull_request:45476" ], - "prepared_review_unit_hash": "0fcb5ee7191237b622e4818eee7ab0463fb3fa20503751318bf1056e36d4734c", + "prepared_review_unit_hash": "37be47827d08eb8738a1bb676fff354d941e4ea415dea2086a19c2a47569ad4b", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7467, - "estimated_input_tokens": 1867, - "estimated_eval_tokens": 3990 + "serialized_chars": 7469, + "estimated_input_tokens": 1868, + "estimated_eval_tokens": 3992 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7467, - "estimated_input_tokens": 1867, - "estimated_eval_tokens": 3990 + "serialized_chars": 7469, + "estimated_input_tokens": 1868, + "estimated_eval_tokens": 3992 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three pull requests appear to be unrelated synthetic/test PRs with no shared underlying bug or fix; the only overlap is incidental, especially the shared vit test file.", - "confidence": 0.81, + "summary": "These are three unrelated '[don't merge]' demo PRs with only incidental overlap in a shared test file; they do not look like duplicates or the same underlying fix.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "43532 is the most representative anchor for the cluster because it is the broadest infra-oriented PR, but it is still just a prank/test change rather than a true duplicate target.", + "canonical_pr_reason": "PR #43532 is the best representative of the cluster because it is the broadest example PR, with the most files and workflow-related changes, while the others are narrower demo changes.", "best_issue_reason": null, - "best_pr_reason": "43532 is the least narrow and most cluster-central of the three, touching CI/testing infrastructure in a way that best represents the group, though none are suitable real merge candidates.", + "best_pr_reason": "PR #43532 is the most representative open PR here: it has the largest surface area and the clearest 'show diff' purpose, making it the best cluster anchor despite not matching the others semantically.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "Both are clearly non-functional test/bot-check PRs, but they edit different areas for different stunts; the shared vit test file is incidental and not the same underlying change." + "reason": "Shared only the vit test file; one PR is a setup/version/dependency tweak, the other is workflow/testing-utils demo changes. Different code paths and purposes." }, { "left": "pull_request:43488", "right": "pull_request:45476", "accept": false, - "reason": "43488 is about setup/version/dependency fiddling, while 45476 adds a workflow caller and dummy tokenization/test edits; they do not fix the same code path or problem." + "reason": "Only incidental overlap on a test file. #43488 is packaging/version/dependency related; #45476 is a CI-workflow call plus unrelated tokenization/test edits." }, { "left": "pull_request:43532", "right": "pull_request:45476", "accept": false, - "reason": "They both touch CI/test infrastructure, but the concrete changes are unrelated and would not plausibly be merged as one fix; shared filenames do not imply duplicate intent." + "reason": "Both touch CI/test areas, but the concrete edits are different: workflow/testing-utils versus a workflow-caller plus clip/chinese_clip test and tokenizer stubs. Not the same change." } ] }, @@ -15763,7 +15755,7 @@ { "label": "pull_request", "component_index": 11, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15779,7 +15771,7 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "6fdc53f98cda7256209b00070c0008d3e83c35e3a38dbe731d69bb50b5cf640e", + "prepared_review_unit_hash": "50d5c410d340e7f730b38bad6591181a007e3e897b1946062a59703f75cc1a53", "component_budget": { "node_count": 2, "item_count": 2, @@ -15801,18 +15793,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate open PRs for different pipeline additions: promptable concept segmentation vs promptable visual segmentation. They share documentation/registry scaffolding, but they target different tasks and model integrations.", - "confidence": 0.97, + "summary": "These are related but distinct PRs: one adds a Promptable Concept Segmentation pipeline, the other adds a Promptable Visual Segmentation pipeline. The overlap is mostly shared registration/docs scaffolding; the actual targets, pipeline names, and model/test files differ, so they should not be deduplicated.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: the two PRs implement different features for different segmentation tasks, so they are not duplicates.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "If a representative PR is needed, #43613 is the broader change set, but it should not subsume #43612 because the underlying feature is different.", + "best_pr_reason": "PR 43613 is the stronger representative of the cluster because it includes broader concrete implementation and test coverage across multiple segmentation model families, making it the more complete standalone change.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "Different concrete features and code paths: concept segmentation vs visual segmentation. Shared docs/registration files are just pipeline boilerplate, not evidence of the same underlying change." + "reason": "Different pipeline features with different task names and model support; shared docs/registration code is generic scaffolding, not evidence of the same underlying change." } ] }, @@ -15826,7 +15818,7 @@ { "label": "pull_request", "component_index": 12, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15842,7 +15834,7 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "0dd466f2584f2e9146a63db270e6505cc0904858aa3ab106eb2db6cd6f5efb77", + "prepared_review_unit_hash": "08af0f941814e2205b1c5d106a540c0fd98b05df8352d43c45273b45a800f28d", "component_budget": { "node_count": 2, "item_count": 2, @@ -15864,18 +15856,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are only loosely related by touching `trainer.py` and adding custom-trainer hooks, but they address different behaviors: gradient-accumulation loss scaling vs. input column filtering. They are not duplicates and should not be merged together.", - "confidence": 0.98, + "summary": "Both pull requests add small extensibility hooks to `Trainer`, but they address different code paths: gradient-accumulation loss scaling vs. dataset column/signature filtering. They are not duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: #43651 changes training-step loss normalization for gradient accumulation, while #43663 refactors signature-based column filtering. They target different code paths and separate behaviors.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "#43651 is the more representative change if a single PR must be chosen, because it directly alters core training loss handling; however, it is not a substitute for #43663.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Reject: both touch `trainer.py`, but one adds `_loss_is_scaled_for_ga` for loss scaling and the other adds `_get_signature_columns` for column filtering. Different concrete behaviors and not plausibly one merged PR." + "reason": "Different underlying changes in `trainer.py`: one alters training-step loss scaling, the other extracts signature-column inspection for column filtering. They are related only at the subsystem level, not the same fix." } ] }, @@ -15889,7 +15881,7 @@ { "label": "pull_request", "component_index": 13, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -15907,52 +15899,52 @@ "pull_request:43656|pull_request:43842", "pull_request:43836|pull_request:43842" ], - "prepared_review_unit_hash": "4c983df41d6ee70b12304ab629cccfca50ecdfa7443b62496eceb6bb547ccf32", + "prepared_review_unit_hash": "7a91a46c8f5292986da50d6dd5e1711474a63c04818142f29065b4bbe4508c3b", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 5992, - "estimated_input_tokens": 1498, - "estimated_eval_tokens": 3252 + "serialized_chars": 5993, + "estimated_input_tokens": 1499, + "estimated_eval_tokens": 3254 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 5992, - "estimated_input_tokens": 1498, - "estimated_eval_tokens": 3252 + "serialized_chars": 5993, + "estimated_input_tokens": 1499, + "estimated_eval_tokens": 3254 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three PRs target the same runtime annotation problem in `src/transformers/cli/serve.py` around `TypeAdapter`/type-hint evaluation. PR 43842 is the clearest minimal fix; the others are broader variants of the same underlying bug.", - "confidence": 0.89, + "summary": "All three PRs target the same TypeAdapter annotation/import problem in `src/transformers/cli/serve.py`. PR 43842 is the most focused fix; 43656 and 43836 look like earlier or broader variants of the same underlying NameError issue.", + "confidence": 0.86, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43842 is the most direct and narrowly-scoped fix: it imports `TypeAdapter` only under `TYPE_CHECKING` and quotes the annotation where it is used, matching the reported NameError without extra refactoring.", + "canonical_pr_reason": "PR 43842 is the cleanest and most specific fix: it directly addresses the `TypeAdapter` NameError when pydantic is absent, with minimal surface area.", "best_issue_reason": null, - "best_pr_reason": "PR 43842 best matches the concrete failure mode and has the smallest, most targeted change set for the `TypeAdapter` NameError in the CLI.", + "best_pr_reason": "PR 43842 is the best representative because it targets the exact runtime failure with the smallest, most explicit change set.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": true, - "reason": "Both modify `src/transformers/cli/serve.py` to prevent `TypeAdapter`/annotation NameErrors at import time; they address the same underlying runtime typing bug and are mergeable as one fix." + "reason": "Both appear to fix the same `TypeAdapter` annotation NameError in `cli/serve.py`; 43656 does it via future annotations/type-hint cleanup, while 43842 uses a more explicit TYPE_CHECKING import/quoted annotation. Same underlying bug, mergeable into one fix." }, { "left": "pull_request:43836", "right": "pull_request:43842", "accept": true, - "reason": "Both are workarounds for the same `TypeAdapter` NameError in the same code path (`cli/serve.py`), with overlapping annotation-stringification/import-guard changes." + "reason": "Both are about the same `TypeAdapter` NameError in the CLI serve path and touch the same code locations. 43836 is broader, but the concrete bug being fixed is the same and could plausibly be handled in one PR." } ] }, "evaluator_result": { "accept": true, - "feedback": "The analyst summary stays within the packet: all three PRs touch `src/transformers/cli/serve.py` and the titles/snippets support a shared TypeAdapter/annotation NameError theme. The soft-edge pairings are also reasonably conservative: they describe overlapping fixes in the same file and do not claim anything stronger than mergeability for the same bugfix." + "feedback": "The summary is grounded in the packet, and both soft-edge accepts are conservative enough: 43836 and 43842 clearly overlap on the same serve.py TypeAdapter fix, and 43656 is at least plausibly the same runtime-annotation bug in the same file. No evidence of an overstatement that would justify rejecting." }, "evaluator_used": true, "retried": false, @@ -15963,151 +15955,144 @@ { "label": "pull_request", "component_index": 14, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", "reason": null, - "source": "llm", - "cache_hit": false, + "source": "cache", + "cache_hit": true, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-43757-3", + "cluster_id": "cluster-43779-2", "nodes": [ - "pull_request:43757", - "pull_request:45500", - "pull_request:45506" + "pull_request:43779", + "pull_request:43816" ], "soft_pairs": [ - "pull_request:45500|pull_request:45506", - "pull_request:43757|pull_request:45500", - "pull_request:43757|pull_request:45506" + "pull_request:43779|pull_request:43816" ], - "prepared_review_unit_hash": "57a95e3b77932958bd01666855a7c41d608d0a32a63b24306a182ec38bdc357f", + "prepared_review_unit_hash": "e7c674adcc031aa8883109a5dfc678b4407fb20b10764302092e384d031ba83a", "component_budget": { - "node_count": 3, - "item_count": 3, - "soft_pair_count": 3, - "serialized_chars": 7755, - "estimated_input_tokens": 1939, - "estimated_eval_tokens": 4134 + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 }, "budget": { - "node_count": 3, - "item_count": 3, - "soft_pair_count": 3, - "serialized_chars": 7755, - "estimated_input_tokens": 1939, - "estimated_eval_tokens": 4134 + "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": "The two GPT-OSS GGUF support PRs are the same end goal, with 45506 being the fuller/latest implementation and the best representative. The earlier PR 43757 is only a fallback workaround and is not the same concrete change as the full-loading PRs.", - "confidence": 0.93, + "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 45506 is the most complete and latest GPT-OSS GGUF support change: it adds the integration mapping, loader support, and tests, and its title explicitly says it supersedes 43757.", + "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 45506 is the best single representative because it fully addresses the GPT-OSS GGUF loading gap and is clearly intended as the final implementation rather than a stopgap.", + "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:45500", - "right": "pull_request:45506", + "left": "pull_request:43779", + "right": "pull_request:43816", "accept": true, - "reason": "Both PRs target the same GPT-OSS GGUF loading feature/fix, share the same issue, filenames, and code path, and 45506 is a direct fuller version of 45500." - }, - { - "left": "pull_request:43757", - "right": "pull_request:45500", - "accept": false, - "reason": "43757 is only a fallback from gpt-oss to gpt-neox, while 45500 aims for full GGUF loading support; they address related but different levels of functionality and are not the same concrete change." - }, - { - "left": "pull_request:43757", - "right": "pull_request:45506", - "accept": false, - "reason": "43757 is a narrow workaround, whereas 45506 implements the actual GPT-OSS GGUF support stack; these are not merge-equivalent fixes for the same underlying change." + "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": null, - "evaluator_used": false, + "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": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, { "label": "pull_request", "component_index": 15, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", "reason": null, - "source": "cache", - "cache_hit": true, + "source": "llm", + "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-43779-2", + "cluster_id": "cluster-43838-3", "nodes": [ - "pull_request:43779", - "pull_request:43816" + "pull_request:43838", + "pull_request:43973", + "pull_request:45283" ], "soft_pairs": [ - "pull_request:43779|pull_request:43816" + "pull_request:43838|pull_request:45283", + "pull_request:43973|pull_request:45283" ], - "prepared_review_unit_hash": "e7c674adcc031aa8883109a5dfc678b4407fb20b10764302092e384d031ba83a", + "prepared_review_unit_hash": "a577666818cc64ae8e846b3dd2c6891718952d9bcd7f6c422d518f83a0ae1523", "component_budget": { - "node_count": 2, - "item_count": 2, - "soft_pair_count": 1, - "serialized_chars": 4362, - "estimated_input_tokens": 1091, - "estimated_eval_tokens": 2438 + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 8697, + "estimated_input_tokens": 2175, + "estimated_eval_tokens": 4606 }, "budget": { - "node_count": 2, - "item_count": 2, - "soft_pair_count": 1, - "serialized_chars": 4362, - "estimated_input_tokens": 1091, - "estimated_eval_tokens": 2438 + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 8697, + "estimated_input_tokens": 2175, + "estimated_eval_tokens": 4606 }, "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, + "analyst_result": { + "summary": "These PRs are related only by broad Qwen/audio metadata, not by the same underlying change. 43838 is the strongest standalone PR here; 45283 is a separate GGUF-loading mapping update, and 43973 is another distinct audio model addition. None of the soft pairs should be merged as duplicates.", + "confidence": 0.89, "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.", + "canonical_pr_reason": "PR 43838 is the most mature and self-contained candidate: it has an explicit issue target, much more discussion/review activity, and a full implementation/test/docs footprint.", "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.", + "best_pr_reason": "PR 43838 is the best representative for the cluster because it is the most complete, most reviewed, and most clearly scoped model-support addition; the other PRs are separate feature tracks.", "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." + "left": "pull_request:43838", + "right": "pull_request:45283", + "accept": false, + "reason": "43838 adds Qwen3-ASR model support and docs; 45283 only adds Qwen3.5 GGUF metadata/mapping support. They touch different code paths and could not plausibly be merged into one PR." + }, + { + "left": "pull_request:43973", + "right": "pull_request:45283", + "accept": false, + "reason": "43973 adds Lfm2Audio model support, while 45283 is a GGUF-loading integration change for Qwen3.5. Same broad audio/modeling area, but not the same bug or change." } ] }, - "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, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, { "label": "pull_request", "component_index": 16, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 2, "status": "reviewed", @@ -16170,8 +16155,8 @@ "pull_request:44018|pull_request:44116", "pull_request:44018|pull_request:44086", "pull_request:44018|pull_request:44085", - "pull_request:44071|pull_request:44085", "pull_request:44068|pull_request:44086", + "pull_request:44071|pull_request:44085", "pull_request:44059|pull_request:44086", "pull_request:44059|pull_request:44722", "pull_request:44085|pull_request:44086", @@ -16191,7 +16176,7 @@ "pull_request:44056|pull_request:44101", "pull_request:44073|pull_request:44161", "pull_request:44101|pull_request:44161", - "pull_request:43997|pull_request:44017", + "pull_request:44017|pull_request:44018", "pull_request:43997|pull_request:44056", "pull_request:44071|pull_request:44076", "pull_request:44074|pull_request:44101", @@ -16220,387 +16205,402 @@ "pull_request:44044|pull_request:44154", "pull_request:44024|pull_request:44076" ], - "prepared_review_unit_hash": "e8b92abd9ae7a81c7f388c370fe76f3ee8b22a98400fc5c06db60fd0c4f58481", + "prepared_review_unit_hash": "675923cc03cb926bfc0681b295d1d1d5f3958ccfc2817c163fee1432f6876ce0", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84501, - "estimated_input_tokens": 21126, - "estimated_eval_tokens": 42508 + "serialized_chars": 84505, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 }, "budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 72, - "serialized_chars": 76934, - "estimated_input_tokens": 19234, - "estimated_eval_tokens": 38724 + "serialized_chars": 76938, + "estimated_input_tokens": 19235, + "estimated_eval_tokens": 38726 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a broad wave of standardized output-tracing refactors across unrelated model families. The only strong duplicate candidates are the GPT-Neo pair and the GPT-J/GPTJ-derived pair(s); the rest are same-theme but different code paths and should stay separate.", - "confidence": 0.86, + "summary": "This cluster is mostly a family of model-specific refactors to standardized output tracing/decorator-based return handling. Most PRs are related only by the umbrella output-tracing effort, not by the same concrete code change. The only strong near-duplicate is the GPT-J/CodeGen refactor pair.", + "confidence": 0.72, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 44722 is the strongest representative: it is an explicit output-tracing refactor with the shared 43979 target, touches the concrete GPT-J/CodeGen code path, and reads like the most complete version of that specific change.", "best_issue_reason": null, - "best_pr_reason": "44066 is the clearest representative of the cluster: it has the explicit #43979 linkage, a concrete GPT-J/CodeGen output-tracing refactor, and the most direct fit for the standardized decorator migration theme.", + "best_pr_reason": "PR 44722 best captures the cluster\u2019s core change pattern and is the clearest anchor for the GPT-J/CodeGen duplicate candidate set.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Different models and code paths; shared issue theme is not enough." + "reason": "Same general output-tracing theme, but different model scope; not the same concrete change." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "CVT/FNet vs DeBERTa-v2 are separate refactors in different model stacks." + "reason": "Both are standardized output refactors, but they target different models and code paths." }, { "left": "pull_request:44066", "right": "pull_request:44085", - "accept": true, - "reason": "Same GPT-J output-tracing refactor family; 44085 appears to be a narrower slice of the same change." + "accept": false, + "reason": "Very similar theme, but 44066 also changes CodeGen; scope is not clearly the same PR." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "RegNet/ResNet vs EfficientNet are different model implementations." + "reason": "Both are refactors for output tracing, but one is ResNet/RegNet and the other is EfficientNet; not duplicates." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "EfficientNet and GPT-J are unrelated code paths despite both touching output tracing." + "reason": "Different model families and different forward paths; umbrella issue match is not enough." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "GPT-J/CodeGen refactor vs EfficientNet refactor; different models." + "reason": "Unrelated model families; only share the broad output-tracing migration theme." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 and DeBERTa-v2 are unrelated changes." + "reason": "Different backbones and different implementation details; not the same underlying bug/change." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen vs MGP-STR are different concrete refactors." + "reason": "Both are decorator refactors, but they affect different models and code paths." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J/CodeGen and MPT are separate model paths." + "reason": "Same broad refactor theme, but MPT and GPT-J are distinct code paths." }, { "left": "pull_request:44018", "right": "pull_request:44068", - "accept": true, - "reason": "Both target the same GPT-Neo output-tracing migration in the same model file." + "accept": false, + "reason": "Related output-tracing refactors, but GPT-Neo and GPT-J are not the same change." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J/CodeGen vs GPT-Neo; same theme, different code path." + "reason": "GPT-J/CodeGen versus GPT-Neo; shared refactor style is not enough." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo and GPT-J are different model families." + "reason": "Different models and different forward implementations." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet and GPT-J are unrelated model refactors." + "reason": "ResNet versus GPT-J/CodeGen; only the umbrella refactor pattern matches." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR and GPT-J are different code paths." + "reason": "MGP-STR and GPT-J/CodeGen are unrelated model-specific changes." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet vs GPT-J; not the same underlying change." + "reason": "ResNet refactor and GPT-J refactor are distinct code changes." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are unrelated implementations." + "reason": "GPT-Neo and ResNet are different implementations; not a duplicate." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet and MPT are separate model migrations." + "reason": "Different backbone families; same migration style only." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert/XLM are different model paths." + "reason": "GPT-Neo versus Flaubert; not the same bug/change." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo and GPT-J are different underlying changes." + "reason": "GPT-Neo and GPT-J/CodeGen are separate refactors despite similar goals." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT and GPT-J are unrelated refactors." + "reason": "MPT and GPT-J/CodeGen are different model paths." }, { "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo vs Flaubert/XLM; same pattern, different code path." + "reason": "Different models and different output-handling implementations." }, { "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo and MGP-STR are distinct model implementations." + "reason": "GPT-Neo and MGP-STR are unrelated; only the output-tracing theme is shared." }, { "left": "pull_request:44018", "right": "pull_request:44085", "accept": false, - "reason": "GPT-Neo and GPT-J are different refactors." + "reason": "Different models and different scopes." }, { - "left": "pull_request:44071", - "right": "pull_request:44085", + "left": "pull_request:44068", + "right": "pull_request:44086", "accept": false, - "reason": "MPT and GPT-J are separate code paths." + "reason": "GPT-Neo and MGP-STR are distinct code paths." }, { - "left": "pull_request:44068", + "left": "pull_request:44071", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo and MGP-STR are unrelated." + "reason": "MPT and MGP-STR are unrelated refactors." }, { - "left": "pull_request:44059", - "right": "pull_request:44086", + "left": "pull_request:43997", + "right": "pull_request:44073", "accept": false, - "reason": "GPT2 and MGP-STR are different model migrations." + "reason": "RegNet and VisualBert are different models; not the same change." }, { "left": "pull_request:44059", - "right": "pull_request:44722", + "right": "pull_request:44071", "accept": false, - "reason": "GPT2 and GPT-J are not the same underlying bug or change." + "reason": "GPT-2 and MPT are separate implementations; same refactor theme only." }, { "left": "pull_request:44085", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J and MGP-STR are unrelated code paths." + "reason": "GPT-J output tracing versus MGP-STR output tracing; different code paths." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT and Flaubert/XLM are different model stacks." - }, - { - "left": "pull_request:44018", - "right": "pull_request:44071", - "accept": false, - "reason": "GPT-Neo and MPT are separate refactors." + "reason": "ViLT and Flaubert are unrelated model-specific refactors." }, { - "left": "pull_request:44017", - "right": "pull_request:44722", + "left": "pull_request:44073", + "right": "pull_request:44161", "accept": false, - "reason": "SegFormer and GPT-J are unrelated model changes." + "reason": "VisualBert and LongT5 are different models; not duplicates." }, { - "left": "pull_request:44017", + "left": "pull_request:44101", "right": "pull_request:44161", "accept": false, - "reason": "SegFormer and LongT5 are different code paths." + "reason": "Flaubert and LongT5 target different architectures and return paths." }, { "left": "pull_request:44017", - "right": "pull_request:44073", - "accept": false, - "reason": "SegFormer and VisualBert are separate model migrations." - }, - { - "left": "pull_request:44068", - "right": "pull_request:44161", + "right": "pull_request:44018", "accept": false, - "reason": "GPT-Neo and LongT5 are different underlying changes." + "reason": "SegFormer and GPT-Neo are unrelated changes." }, { "left": "pull_request:43997", - "right": "pull_request:44073", + "right": "pull_request:44056", "accept": false, - "reason": "RegNet and VisualBert are unrelated refactors." + "reason": "RegNet and MPNet are different models; no shared concrete bug." }, { - "left": "pull_request:43997", - "right": "pull_request:44056", + "left": "pull_request:44071", + "right": "pull_request:44076", "accept": false, - "reason": "RegNet and MPNet are separate model paths." + "reason": "MPT and ImageGPT are separate output-tracing refactors." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet and Flaubert/XLM are different code paths." + "reason": "TextNet and Flaubert are different code paths." }, { "left": "pull_request:44056", "right": "pull_request:44161", "accept": false, - "reason": "MPNet and LongT5 are unrelated migrations." + "reason": "MPNet and LongT5 are unrelated beyond the generic refactor theme." }, { "left": "pull_request:43997", "right": "pull_request:44018", "accept": false, - "reason": "RegNet and GPT-Neo are different model families." + "reason": "RegNet and GPT-Neo are different model implementations." }, { - "left": "pull_request:44071", + "left": "pull_request:43997", "right": "pull_request:44161", "accept": false, - "reason": "MPT and LongT5 are separate refactors." + "reason": "RegNet and LongT5 are unrelated." }, { - "left": "pull_request:43997", - "right": "pull_request:44017", + "left": "pull_request:44074", + "right": "pull_request:44161", "accept": false, - "reason": "RegNet and SegFormer are different code paths." + "reason": "TextNet and LongT5 do not share the same underlying change." }, { - "left": "pull_request:44010", + "left": "pull_request:43997", "right": "pull_request:44071", "accept": false, - "reason": "SqueezeBert and MPT are unrelated changes." + "reason": "RegNet and MPT are different backbones; same umbrella issue is insufficient." }, { - "left": "pull_request:44044", - "right": "pull_request:44074", + "left": "pull_request:44010", + "right": "pull_request:44071", "accept": false, - "reason": "DeBERTa-v2 and TextNet are distinct model implementations." + "reason": "SqueezeBert and MPT are separate model-specific refactors." }, { "left": "pull_request:44010", "right": "pull_request:44073", "accept": false, - "reason": "SqueezeBert and VisualBert are different underlying changes." + "reason": "SqueezeBert and VisualBert are different architectures and code paths." }, { - "left": "pull_request:44010", - "right": "pull_request:44044", + "left": "pull_request:44018", + "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert and DeBERTa-v2 are separate refactors." + "reason": "GPT-Neo and SpeechT5 are unrelated model families." }, { - "left": "pull_request:44044", + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different implementations." + }, + { + "left": "pull_request:43997", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa-v2 and SpeechT5 are unrelated model code paths." + "reason": "RegNet and SpeechT5 are not the same underlying change." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "SqueezeBert and GPT-Neo are separate refactors." }, { "left": "pull_request:44071", - "right": "pull_request:44076", + "right": "pull_request:44129", "accept": false, - "reason": "MPT and ImageGPT are different models." + "reason": "MPT and SpeechT5 have different forward/output paths." }, { - "left": "pull_request:44074", + "left": "pull_request:43996", "right": "pull_request:44161", "accept": false, - "reason": "TextNet and LongT5 are unrelated." + "reason": "CVT/FNet and LongT5 are unrelated beyond the refactor pattern." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "FNet/CVT and SegFormer are different model-specific changes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBert and DeBERTa-v2 are not the same code-path fix." }, { "left": "pull_request:44044", - "right": "pull_request:44154", + "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa-v2 and VITS are separate model migrations." + "reason": "DeBERTa-v2 and TextNet are separate output-tracing refactors." }, { - "left": "pull_request:44024", - "right": "pull_request:44076", + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor on the same code path; 44722 looks like a later rebased/expanded version of the same change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", "accept": false, - "reason": "FocalNet and ImageGPT are different concrete refactors." + "reason": "Very similar topic, but 44085 is GPT-J-only while 44722 also includes CodeGen; not clearly the same exact change." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J and SpeechT5 are unrelated code paths." + "reason": "GPT-J output tracing versus SpeechT5 output tracing are unrelated." }, { "left": "pull_request:43996", "right": "pull_request:44007", "accept": false, - "reason": "CVT/FNet and RegNet/ResNet/RT-DETR are different model migrations." + "reason": "CVT/FNet and ResNet/RegNet are different model families; no duplicate evidence." }, { "left": "pull_request:43996", "right": "pull_request:44129", "accept": false, - "reason": "CVT/FNet and SpeechT5 are unrelated." + "reason": "FNet/CVT and SpeechT5 are unrelated." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa-v2 and SpeechT5 are different underlying changes." + "reason": "DeBERTa-v2 and SpeechT5 are separate model-specific refactors." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa-v2 and VITS are separate model refactors." + "reason": "DeBERTa-v2 and VITS are different code paths and architectures." }, { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "FocalNet and ImageGPT are unrelated implementations." + "reason": "FocalNet and ImageGPT are unrelated models; same refactor theme only." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it identifies the main GPT-Neo and GPT-J/CodeGen duplicate candidates without overextending to the unrelated model refactors. The accepted soft-edge pairs are plausible as mergeable same-bugfix changes within the same code paths, and the rejected pairs are appropriately cautious." - }, - "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 }, { "label": "pull_request", "component_index": 16, - "component_count": 22, + "component_count": 24, "review_unit_index": 2, "review_unit_count": 2, "status": "reviewed", @@ -16624,14 +16624,14 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "747e937f931315c7a3892b707ed3505d52bf30bb0f2791999e13a3ae6cdf650a", + "prepared_review_unit_hash": "ef6a9f020cda65bc8a8cccc2827232d9d7a6745c8804f9843d2c59f8c3a80b42", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84501, - "estimated_input_tokens": 21126, - "estimated_eval_tokens": 42508 + "serialized_chars": 84505, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 }, "budget": { "node_count": 6, @@ -16646,42 +16646,42 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These are related model-by-model refactors toward standardized output tracing, but they are not duplicates of the same concrete change. Each PR targets a different architecture/code path, so they should stay separate.", + "summary": "These PRs are all separate model-by-model refactors for standardized output tracing/capture, not duplicates of the same code change. They share the same migration theme but touch different model implementations and code paths.", "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44129 is the best anchor only because it has an explicit issue target and a clearly scoped standalone refactor; however, it does not subsume the other model-specific changes.", + "canonical_pr_reason": "No single canonical duplicate PR stands out; the closest representative is 44073 because it cleanly wires in capture_outputs and _can_record_outputs for one model, but the other PRs are still distinct model-specific changes.", "best_issue_reason": null, - "best_pr_reason": "44129 is the strongest representative of the cluster due to its explicit issue linkage and substantial output-tracing refactor, but the rest are still distinct per-model implementations.", + "best_pr_reason": "44073 is the best representative of the cluster\u2019s theme: a focused, self-contained standardized output-tracing refactor with explicit model output-recording setup.", "soft_edge_verdicts": [ { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Same refactor theme, but different models and different implementation surfaces; not the same underlying bug/change." + "reason": "Both are output-tracing refactors, but for different models (VisualBert vs TextNet) with different code paths; same migration pattern, not the same underlying change." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "Both modernize output tracing, but for different model code paths (MPNet vs VisualBert), so they are not duplicates." + "reason": "Both add standardized output capture, but they target different models (MPNet vs VisualBert) and different model internals; not mergeable as one duplicate PR." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "Shared decorator/output-capture pattern only; the actual changes are model-specific and not mergeable as one concrete fix." + "reason": "Same refactor family, but distinct model implementations (MPNet vs TextNet) and separate forward/output wiring; not the same concrete bug or change." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert are separate implementations of the same style of cleanup, not the same underlying PR." + "reason": "RegNet and SqueezeBert are unrelated model code paths; these are parallel refactors, not duplicate fixes." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "Both touch standardized output handling, but they affect different model internals and do not fix the same concrete code path." + "reason": "SqueezeBert output-collection refactor and SpeechT5 output-tracing refactor are different model-specific changes despite similar terminology." } ] }, @@ -16695,7 +16695,7 @@ { "label": "pull_request", "component_index": 17, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -16732,16 +16732,13 @@ "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:44000|pull_request:44026", "pull_request:44001|pull_request:44002", + "pull_request:44001|pull_request:44003", "pull_request:44001|pull_request:44025", - "pull_request:44001|pull_request:44026", "pull_request:44002|pull_request:44029", - "pull_request:44002|pull_request:44025", - "pull_request:44001|pull_request:44003", "pull_request:44002|pull_request:44003", - "pull_request:44000|pull_request:44003", "pull_request:44003|pull_request:44004", "pull_request:43998|pull_request:44004", "pull_request:44001|pull_request:44004", @@ -16763,286 +16760,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "5f9f9b6b3946c6fa3a69370e0fe383b8baa69acc7c191417d0f701c9046748c9", + "prepared_review_unit_hash": "e39351b5510747e009cf6432229fb40128fe942d1722c5dc313c88a77262b824", "component_budget": { "node_count": 13, "item_count": 13, - "soft_pair_count": 42, - "serialized_chars": 40933, - "estimated_input_tokens": 10234, - "estimated_eval_tokens": 20724 + "soft_pair_count": 39, + "serialized_chars": 39780, + "estimated_input_tokens": 9945, + "estimated_eval_tokens": 20146 }, "budget": { "node_count": 13, "item_count": 13, - "soft_pair_count": 42, - "serialized_chars": 40933, - "estimated_input_tokens": 10234, - "estimated_eval_tokens": 20724 + "soft_pair_count": 39, + "serialized_chars": 39780, + "estimated_input_tokens": 9945, + "estimated_eval_tokens": 20146 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are independent model-specific output-tracing refactors across different architectures, not duplicate bug fixes or the same concrete change.", - "confidence": 0.95, + "summary": "This cluster is a family of independent PRs that all refactor output tracing / `can_return_tuple` handling across different model implementations. They share the same theme, but each targets a different concrete code path, so they should not be merged as duplicates.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44003 is the best representative: it has the broadest refactor footprint, touches two related Mamba files, and includes the clearest output-capturing/tracing changes (`can_return_tuple`, `capture_outputs`, `_can_record_outputs`).", + "canonical_pr_reason": "No issues are present; among the PRs, #44003 is the best representative because it touches two closely related Mamba implementations and shows the shared output-capturing refactor most clearly.", "best_issue_reason": null, - "best_pr_reason": "PR 44003 best captures the cluster\u2019s main theme while being the most substantial example of the output-tracing refactor pattern.", + "best_pr_reason": "#44003 is the strongest representative of the cluster: it is a more substantial refactor, spans related model files, and best exemplifies the shared output-tracing change pattern.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Same refactor theme, but different models (`timm_backbone` vs `upernet`) and unrelated code paths." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Both are output-tracing refactors, but they affect different model families (`timm_backbone` vs `vision_text_dual_encoder`)." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Shared output-tracing cleanup, but `timm_backbone` and `univnet` are unrelated implementations." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Different model stacks (`timm_backbone` vs Mamba/FalconMamba); same pattern, not the same bug/change." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Both refactor tracing, but `timm_backbone` and `rwkv` are separate code paths with no shared concrete fix." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Similar tracing refactor, but `mobilenet_v1` and `univnet` are different models and changes are not mergeable as one PR." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`mobilenet_v1` and `upernet` share a theme only; they fix different model-specific forward paths." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Both are tracing refactors, but they apply to different depth-related models and distinct code paths." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`mobilenet_v1` and `vision_encoder_decoder` are unrelated architectures; same refactor pattern is not enough." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Different model families (`mobilenet_v1` vs `speech_encoder_decoder`), so this is not one underlying change." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Both modernize output handling, but `vision_text_dual_encoder` and `univnet` are separate concrete implementations." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Same broad refactor style, but `vision_text_dual_encoder` and `upernet` do not share a single code-path fix." }, { "left": "pull_request:44000", - "right": "pull_request:44025", + "right": "pull_request:44003", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`vision_text_dual_encoder` vs Mamba/FalconMamba: different subsystems, so not duplicates." }, { "left": "pull_request:44000", - "right": "pull_request:44026", + "right": "pull_request:44025", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Same output-tracing theme, but these are separate model refactors with different forward signatures." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`univnet` and `upernet` are unrelated; the overlap is only in output-tracing cleanup." }, { "left": "pull_request:44001", - "right": "pull_request:44025", + "right": "pull_request:44003", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Different model code paths (`univnet` vs Mamba/FalconMamba); not the same bug or fix." }, { "left": "pull_request:44001", - "right": "pull_request:44026", - "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." - }, - { - "left": "pull_request:44002", - "right": "pull_request:44029", - "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." - }, - { - "left": "pull_request:44002", "right": "pull_request:44025", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Both are tracing refactors, but `univnet` and `depth_anything` are distinct model implementations." }, { "left": "pull_request:44001", - "right": "pull_request:44003", + "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`univnet` and `codegen` target different forward logic; shared theme only, no duplicate change." }, { "left": "pull_request:44002", - "right": "pull_request:44003", + "right": "pull_request:44029", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`upernet` and `rwkv` are unrelated model paths; the PRs are not the same concrete fix." }, { - "left": "pull_request:44000", + "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Different model families (`upernet` vs Mamba/FalconMamba), so these are separate refactors." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Mamba/FalconMamba output tracing vs `codegen` cache/output handling are different code paths." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`timm_backbone` and `codegen` are unrelated implementations; same refactor pattern does not make them duplicates." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`univnet` and `codegen` differ in model architecture and forward semantics." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`upernet` and `codegen` touch different subsystems and are not one mergeable change." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`vision_text_dual_encoder` and `codegen` are separate code paths; only the refactor style matches." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "Both are output-tracing refactors, but `depth_anything` and `vision_encoder_decoder` are different model families." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`depth_anything` and `speech_encoder_decoder` are unrelated implementations, so not the same underlying change." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`depth_anything` vs `superpoint`: different model code paths with only thematic overlap." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`depth_anything` and `rwkv` are unrelated; same output-tracing cleanup pattern only." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`depth_anything` and `dpr` target different architectures and cannot be merged as one PR." }, { "left": "pull_request:44026", "right": "pull_request:44029", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`vision_encoder_decoder` and `rwkv` share only the tracing-refactor theme, not a concrete bug fix." }, { "left": "pull_request:44026", "right": "pull_request:44027", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`vision_encoder_decoder` and `speech_encoder_decoder` are related in naming, but still separate code paths and PRs." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`vision_encoder_decoder` and `superpoint` are different model implementations, so this is not duplicate work." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`vision_encoder_decoder` and `dpr` are distinct forward paths; same general refactor style only." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`speech_encoder_decoder` and `superpoint` are unrelated model changes, despite similar output-tracing cleanup." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`speech_encoder_decoder` and `rwkv` affect different code paths and are not the same concrete change." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`speech_encoder_decoder` and `dpr` are separate models; no shared duplicate bug is evident." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`superpoint` and `rwkv` are unrelated implementations; same refactor theme is insufficient." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`superpoint` and `dpr` are different code paths, so they should remain separate PRs." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations/files; same refactor pattern only." + "reason": "`rwkv` and `dpr` are distinct model changes with no shared concrete fix." } ] }, @@ -17056,7 +17041,7 @@ { "label": "pull_request", "component_index": 18, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -17075,7 +17060,7 @@ "pull_request:44662|pull_request:45075", "pull_request:45075|pull_request:45401" ], - "prepared_review_unit_hash": "1314b1454bc4fe600085b173a662664af24d833b409d50ee94c353ee455ec5ad", + "prepared_review_unit_hash": "b968fb6bc2fe42f3ed81c540ecec7177b3978c1b75a1d2c84ef7dff5e2b1ef2b", "component_budget": { "node_count": 3, "item_count": 3, @@ -17097,30 +17082,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three separate model-addition PRs for different architectures (PenguinVL, DeepSeek-OCR-2, Voxtral TTS). Their overlap is mostly shared Transformers boilerplate (auto registries, docs TOC, tests), not the same underlying change.", - "confidence": 0.96, + "summary": "These are three separate model-addition PRs for different architectures (PenguinVL, DeepSeek-OCR-2, Voxtral TTS). The shared file patterns are just Transformers boilerplate, not the same underlying change.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: each PR introduces a different model and codepath, so they should not be merged as duplicates.", + "canonical_pr_reason": "No canonical duplicate PR here: each PR implements a different model and code path, so they should stay separate.", "best_issue_reason": null, - "best_pr_reason": "PR 45401 is the most self-contained representative of its own feature work because it has an explicit tracking issue and a focused model implementation, but it is still not a duplicate of the other two.", + "best_pr_reason": "PR 45075 is the strongest standalone candidate: it is non-draft, has a full model integration including conversion tooling and tests, and appears more complete than the draft Voxtral TTS PR.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Different model integrations (PenguinVL vs Voxtral TTS); shared files are generic registry/docs plumbing, not the same concrete change." + "reason": "Different models and tasks (PenguinVL vs Voxtral TTS); the overlap is only in shared Transformers scaffolding and docs/auto-registry files." }, { "left": "pull_request:44662", "right": "pull_request:45075", "accept": false, - "reason": "Different model families and implementations; overlap is limited to common Transformers scaffolding and does not indicate one underlying bug or feature." + "reason": "Both add unrelated model implementations for different architectures; shared filenames are generic integration points, not evidence of the same bug or change." }, { "left": "pull_request:45075", "right": "pull_request:45401", "accept": false, - "reason": "DeepSeek-OCR-2 and Voxtral TTS are unrelated model additions; the shared auto-registration/docs edits are broad infrastructure, not a mergeable duplicate PR." + "reason": "DeepSeek-OCR-2 and Voxtral TTS are distinct model additions with different code paths and assets; they are not plausibly mergeable into one PR." } ] }, @@ -17134,7 +17119,7 @@ { "label": "pull_request", "component_index": 19, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -17150,7 +17135,7 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "64ce9d5c6049897912ba92401a9c10ddb8d36f690ed590d6472e53ab021d3d75", + "prepared_review_unit_hash": "740406d4aba4a9eef4c26cdcd9008e7364af201a307f74c44f2e12bbd9ce2d47", "component_budget": { "node_count": 2, "item_count": 2, @@ -17172,18 +17157,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related by the broad idea of fusion, but they target different layers and different code paths: PR 44979 introduces a standalone module-fusion API, while PR 45363 wires n-to-1 kernel fusion into KernelConfig/model loading. They should not be treated as duplicates.", - "confidence": 0.88, + "summary": "Both pull requests are about fusion, but they target different layers: one introduces a general Module Fusion API, while the other wires up n-to-1 kernel fusion through KernelConfig and related integration changes. They are thematically related but not the same underlying change.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45363 is the more concrete and integrated implementation: it modifies model loading, integration hooks, and KernelConfig to enable a specific fusion path, whereas PR 44979 is a separate foundational API addition.", + "canonical_pr_reason": "Choose PR 45363 as the representative PR because it spans the actual end-to-end kernel-fusion integration path across core modules, config, and an example, whereas PR 44979 is a narrower standalone API addition.", "best_issue_reason": null, - "best_pr_reason": "PR 45363 is the better representative for this cluster because it addresses an end-to-end user-facing fusion feature and has the broader, more complete change set.", + "best_pr_reason": "PR 45363 is the strongest cluster representative: it is a concrete, multi-file implementation of kernel fusion behavior and is more complete than the standalone Module Fusion API proposal in PR 44979.", "soft_edge_verdicts": [ { - "left": "pull_request:44979", - "right": "pull_request:45363", + "left": "pull_request:44979 (Module Fusion API)", + "right": "pull_request:45363 (n-to-1 kernel fusion via `KernelConfig`)", "accept": false, - "reason": "Different implementations of fusion at different abstraction levels: 44979 adds a generic module fusion API, while 45363 implements n-to-1 kernel fusion via KernelConfig and model integration. They are related conceptually but not the same concrete change." + "reason": "Related fusion infrastructure, but different artifacts and different concrete changes: a new module-fusion API versus kernel-fusion integration and example code. They do not look mergeable as one PR." } ] }, @@ -17197,7 +17182,7 @@ { "label": "pull_request", "component_index": 20, - "component_count": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -17213,7 +17198,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "a25a1cf01d5a3d9ed71f74f0a5529ca4c001ca8e706927323857ddb73383281b", + "prepared_review_unit_hash": "bcadfdf72acceb8dc1cfbea34780e84b56ef5305fe9e4f1c12fb4c06e20a01aa", "component_budget": { "node_count": 2, "item_count": 2, @@ -17235,18 +17220,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests address the same AttributeError in `_patch_mistral_regex` for the same target issue, with nearly identical code changes; PR 45317 is the more complete version because it also adds test coverage.", + "summary": "Both pull requests 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 tests and has clearer context.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45317 is the stronger canonical PR because it fixes the same bug as 45086 and also includes tests, making it the more complete and review-ready change set.", + "canonical_pr_reason": "PR 45317 is the stronger canonical choice: it addresses the same bug as 45086 but includes test coverage and a fuller implementation description, making it the more complete fix.", "best_issue_reason": null, - "best_pr_reason": "PR 45317 is the best PR candidate since it preserves the same fix while adding validation coverage; 45086 appears to be an earlier/smaller variant of the same change.", + "best_pr_reason": "PR 45317 is the best representative of the cluster because it fixes the concrete Mistral pre-tokenizer AttributeError and verifies the behavior with tests.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying AttributeError fix in the same code path, same explicit issue target, and effectively the same patch; the second PR just adds tests." + "reason": "Same underlying bug and same code path in `src/transformers/tokenization_utils_tokenizers.py`; 45317 is essentially an expanded version of 45086 with tests, so these are duplicate fixes that could plausibly be merged into one PR." } ] }, @@ -17260,7 +17245,70 @@ { "label": "pull_request", "component_index": 21, - "component_count": 22, + "component_count": 24, + "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": "2585854094ac5783b4115ee2ded38e437fd8e270b30c53381b74f1b7f65c5aa7", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6137, + "estimated_input_tokens": 1535, + "estimated_eval_tokens": 3326 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6137, + "estimated_input_tokens": 1535, + "estimated_eval_tokens": 3326 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs work on Nemotron-H, but they target different concrete changes: #45115 is a narrower refactor around inheriting GraniteMoeHybrid and conversion/key renaming, while #45569 is a broader model-family split that adds dense/sparse variants, auto-mapping, config updates, and tests. They are related but not duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "#45569 is the better representative PR for this cluster because it is the more complete, newer implementation of the Nemotron-H family split and includes the full set of codepaths and tests, whereas #45115 is a narrower refactor focused mainly on inheritance and checkpoint conversion naming.", + "best_issue_reason": null, + "best_pr_reason": "#45569 is the strongest standalone PR: it covers the actual model/configuration registration work for Nemotron-H dense and sparse variants across auto mappings and tests. #45115 is useful background but is too narrow to serve as the main cluster representative.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Related Nemotron-H work, but not the same underlying bug/change. #45115 is a focused GraniteMoeHybrid-style refactor and conversion-mapping rename, while #45569 introduces separate dense/sparse model variants, config enum updates, auto-class wiring, and tests. These would not 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": 22, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -17276,7 +17324,7 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "2027bd7f06c56bcdaf7cf8ade211e0e2082db64c8cdfaec1a41ec5fa61830b0e", + "prepared_review_unit_hash": "06c45e7a2683c407ff688dd9a6a0c90e4a0ff32effbc9fccd9f68e1b2b04c6ea", "component_budget": { "node_count": 2, "item_count": 2, @@ -17298,18 +17346,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related but distinct PRs: one introduces heterogeneous per-layer config infrastructure, while the other builds on it to add model/cache/masking support. They should not be treated as duplicates.", - "confidence": 0.92, + "summary": "Both PRs target heterogeneous-layer support, but they are not the same change: #45333 is config-only, while #45332 extends the feature into cache/modeling paths and multiple model files. Related, but not duplicate enough to merge as one PR.", + "confidence": 0.86, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45333 is the more foundational, narrower change: it adds the heterogeneous per-layer config machinery that the broader modeling PR depends on.", + "canonical_pr_reason": "#45332 is the broader end-to-end implementation of heterogeneous model support, covering config plus modeling/cache integration, so it is the better representative PR for the cluster.", "best_issue_reason": null, - "best_pr_reason": "PR #45333 is the best canonical PR because it cleanly captures the core enabling feature without the extra model-side plumbing from #45332.", + "best_pr_reason": "#45332 is the more complete and central change; #45333 looks like a narrower prerequisite/subfeature rather than a duplicate of the full implementation.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "Related feature work, but not the same concrete change: #45333 adds per-layer config support, while #45332 extends that into heterogeneous modeling/cache/masking behavior. They would not plausibly be merged as one duplicate PR." + "reason": "They share heterogeneous-config code, but #45333 only adds per-layer config while #45332 also changes cache/modeling logic and model-specific files. Same feature area, not the same concrete change or code path." } ] }, @@ -17322,8 +17370,8 @@ }, { "label": "pull_request", - "component_index": 22, - "component_count": 22, + "component_index": 23, + "component_count": 24, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -17339,7 +17387,7 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "d8ebf51839231c40418f31d25c1fcc396a9f1ad52765c2a3229af6cdde0c4b20", + "prepared_review_unit_hash": "45b9a731de260927adf80c4322c007a33d08ee64acba4a009b3319e61400c0cd", "component_budget": { "node_count": 2, "item_count": 2, @@ -17361,18 +17409,81 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests are typing-related, but they are different changes: one is a broad type-checking sweep across many Transformers modules, while the other adds specific typing protocols/aliases for modeling_utils and PEFT integration.", - "confidence": 0.9, + "summary": "Both PRs are typing-focused maintenance, but they target different scopes: one is a broad type-checking sweep across many transformers modules, while the other adds specific typing support around modeling_utils/PEFT integration. They are related, but not the same underlying change.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45425 is the more concrete, self-contained typing change, centered on new shared typing definitions and modeling_utils/PEFT compatibility.", + "canonical_pr_reason": "PR #45425 is the more focused and self-contained typing change, centered on concrete protocol/alias additions and modeling_utils integration; PR #45415 is a broad repo-wide type-checking pass.", "best_issue_reason": null, - "best_pr_reason": "PR #45425 is the best representative of the cluster if you want the narrowest, most clearly scoped typing-focused change.", + "best_pr_reason": "PR #45425 is the best representative of the cluster because it is narrower and more specific, making it a cleaner exemplar than the wide-ranging type-checking sweep in #45415.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Reject: these are not the same underlying change. #45415 is a broad repository-wide type-checking cleanup/ignore sweep, while #45425 adds specific typing helpers and model-specific protocol support. They share a subsystem, but not a concrete code-path fix." + "reason": "Both are typing-related, but they address different changes: #45415 is a broad type-checking cleanup across many files, while #45425 adds focused typing support for modeling_utils/PEFT. They do not look mergeable as one concrete 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": 24, + "component_count": 24, + "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-45565-2", + "nodes": [ + "pull_request:45565", + "pull_request:45575" + ], + "soft_pairs": [ + "pull_request:45565|pull_request:45575" + ], + "prepared_review_unit_hash": "32f68244295d4371bc1aa86f3721e6c39c2b73a85a273a0be48febffd38798d8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4253, + "estimated_input_tokens": 1064, + "estimated_eval_tokens": 2384 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4253, + "estimated_input_tokens": 1064, + "estimated_eval_tokens": 2384 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs patch the same continuous-batching warning block in `src/transformers/generation/utils.py` and address the stale `num_return_sequences` warning; they look like two variants of the same fix, with 45575 being the narrower direct edit.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "45575 is the cleaner, more directly scoped fix: it removes the stale `num_return_sequences` warning and leaves the remaining `num_beams` check/message aligned with the code path.", + "best_issue_reason": null, + "best_pr_reason": "45575 best represents the cluster because it most precisely matches the underlying warning bug with minimal code churn.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45565", + "right": "pull_request:45575", + "accept": true, + "reason": "Same file, same `generate()` continuous-batching branch, same stale warning problem; the two diffs are close enough that they could plausibly be merged into one PR." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index 03e650e5842ec885dc79622a687302d231a68dbd..7c0e6df6a98965b7d780ee95060571144562fb4c 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260422t060051z", + "analysis_id": "hybrid-model-20260422t120052z", "archived_artifacts": { - "hybrid": "snapshots/20260422T060051Z/analysis-runs/hybrid-model-20260422t060051z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260422T060051Z/analysis-runs/hybrid-model-20260422t060051z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-04-22T06:09:48Z", + "published_at": "2026-04-22T12:09:14Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260422T060051Z", + "snapshot_id": "20260422T120052Z", "variant": "hybrid" } diff --git a/snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.json b/snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..0b2f27ac7682a9cbd282f079e2e26198adf47d3f --- /dev/null +++ b/snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.json @@ -0,0 +1,1656 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260422T120052Z", + "generated_at": "2026-04-22T12:09:12Z", + "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-41211-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41211.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 41211, + "canonical_pr_number": 44339, + "issue_numbers": [ + 41211 + ], + "pr_numbers": [ + 41356, + 44339 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 41356, + "right_pr_number": 44339, + "code_similarity": 0.156, + "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-43656-4", + "summary": "Cluster of 1 issues and 3 PRs centered on issue #43824.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 43824, + "canonical_pr_number": 43656, + "issue_numbers": [ + 43824 + ], + "pr_numbers": [ + 43656, + 43836, + 43842 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43656, + "right_pr_number": 43836, + "code_similarity": 0.36, + "size_similarity": 0.176, + "file_overlap": 1.0, + "area_overlap": 0.051, + "patch_similarity": 0.048, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 + ] + ], + "right_ranges": [ + [ + 11, + 18 + ], + [ + 359, + 429 + ], + [ + 584, + 590 + ], + [ + 1892, + 1910 + ], + [ + 1917, + 1923 + ] + ] + } + ] + }, + { + "left_pr_number": 43656, + "right_pr_number": 43842, + "code_similarity": 0.405, + "size_similarity": 0.5, + "file_overlap": 1.0, + "area_overlap": 0.0, + "patch_similarity": 0.036, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 + ] + ], + "right_ranges": [ + [ + 54, + 61 + ], + [ + 587, + 593 + ] + ] + } + ] + }, + { + "left_pr_number": 43836, + "right_pr_number": 43842, + "code_similarity": 0.332, + "size_similarity": 0.088, + "file_overlap": 1.0, + "area_overlap": 0.033, + "patch_similarity": 0.017, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 359, + 429 + ], + [ + 584, + 590 + ], + [ + 1892, + 1910 + ], + [ + 1917, + 1923 + ] + ], + "right_ranges": [ + [ + 54, + 61 + ], + [ + 587, + 593 + ] + ] + } + ] + } + ] + }, + { + "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-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-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-45563-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45563.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45563, + "canonical_pr_number": 45565, + "issue_numbers": [ + 45563 + ], + "pr_numbers": [ + 45565, + 45575 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45565, + "right_pr_number": 45575, + "code_similarity": 0.709, + "size_similarity": 0.667, + "file_overlap": 1.0, + "area_overlap": 0.526, + "patch_similarity": 0.607, + "shared_filenames": [ + "src/transformers/generation/utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/generation/utils.py", + "left_ranges": [ + [ + 2300, + 2318 + ] + ], + "right_ranges": [ + [ + 2305, + 2314 + ] + ] + } + ] + } + ] + } + ], + "duplicate_issues": [], + "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, shared_issue_target 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-43656-4", + "canonical_pr_number": 43656, + "duplicate_pr_numbers": [ + 43836, + 43842 + ], + "target_issue_number": 43824, + "reason": "PRs in cluster-43656-4 are treated as duplicates because they converge on issue #43824 with closing_reference, shared_issue_target, 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-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-45563-3", + "canonical_pr_number": 45565, + "duplicate_pr_numbers": [ + 45575 + ], + "target_issue_number": 45563, + "reason": "PRs in cluster-45563-3 are treated as duplicates because they converge on issue #45563 with closing_reference, shared_issue_target, soft_similarity evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-41211-3", + "issue_number": 41211, + "reason": "Issue #41211 is the strongest global issue candidate because it is open, belongs to a cluster with 3 artifacts, and carries 6 discussion comments plus 6 inbound references.", + "score": 148.0 + }, + "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/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..0c65014f0cf6cc4bf5af227a6d29f8f07c19a813 --- /dev/null +++ b/snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,17498 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260422T120052Z", + "generated_at": "2026-04-22T12:09:12Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 13, + "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": "cf649f9b9856122366cdb67fd4649209142ca58ef11b06e55c7b69ac3efe0c11", + "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": "The cluster does not look like a single duplicate set. One issue is unrelated (Tapas tokenization), and the two model-loading issues are only loosely similar from their titles alone.", + "confidence": 0.73, + "canonical_issue_reason": "Issue #16618 is the broadest and most general model-loading report in the group, so it is the best candidate to serve as the hub if these were to be grouped.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #16618 is the best representative because its title is generic and could subsume related loading failures, while #15980 is specifically about private-model authentication and #13244 is unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Different underlying problems: TAPAS tokenization mismatch versus a model-loading failure for bert-base-uncased. Same broad ML ecosystem, but not the same bug." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "#15980 is specifically about a bad auth/error message when downloading a private model, while #16618 concerns failing to load bert-base-uncased. Titles suggest different failure modes and likely different fixes." + } + ] + }, + "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": 13, + "review_unit_index": 1, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:24643", + "issue:30277", + "issue:34634", + "issue:34928", + "issue:35545", + "issue:36010", + "issue:36331", + "issue:39290", + "issue:41093", + "issue:41720", + "issue:41762", + "issue:42915", + "issue:43039", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43425", + "issue:43531", + "issue:43541", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43792", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43901", + "issue:43975", + "issue:44292", + "issue:44322", + "issue:44457", + "issue:44509", + "issue:44512", + "issue:44560", + "issue:44661", + "issue:44805", + "issue:44841", + "issue:44863", + "issue:44918", + "issue:45070", + "issue:45081", + "issue:45084", + "issue:45161", + "issue:45237", + "issue:45362", + "issue:45464", + "issue:45507" + ], + "soft_pairs": [ + "issue:44322|issue:45464", + "issue:43716|issue:45237", + "issue:43828|issue:45237", + "issue:45237|issue:45507", + "issue:44292|issue:45237", + "issue:30277|issue:35545", + "issue:35545|issue:42915", + "issue:35545|issue:41720", + "issue:44509|issue:44512", + "issue:35545|issue:36010", + "issue:43638|issue:44805", + "issue:43596|issue:44805", + "issue:44805|issue:45161", + "issue:44805|issue:44918", + "issue:44661|issue:44805", + "issue:34634|issue:35545", + "issue:30277|issue:43638", + "issue:43596|issue:43638", + "issue:43975|issue:44457", + "issue:43039|issue:45161", + "issue:44292|issue:44457", + "issue:34928|issue:44805", + "issue:24643|issue:30277", + "issue:43975|issue:45237", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:39290|issue:45081", + "issue:43828|issue:45161", + "issue:43425|issue:44292", + "issue:44918|issue:45161", + "issue:34928|issue:41720", + "issue:43541|issue:43716", + "issue:43828|issue:44292", + "issue:43716|issue:44292", + "issue:44292|issue:44918", + "issue:43039|issue:44918", + "issue:41762|issue:43638", + "issue:43827|issue:44512", + "issue:43716|issue:44560", + "issue:43716|issue:44918", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:43039|issue:44805", + "issue:34928|issue:36010", + "issue:41762|issue:43596", + "issue:43901|issue:44512", + "issue:39290|issue:41720", + "issue:43866|issue:44863", + "issue:43039|issue:44560", + "issue:43854|issue:43866", + "issue:43425|issue:45070", + "issue:39290|issue:43531", + "issue:41093|issue:43638", + "issue:43404|issue:43866", + "issue:45362|issue:45464", + "issue:43792|issue:45070", + "issue:43541|issue:43828", + "issue:36010|issue:39290", + "issue:44841|issue:45084", + "issue:41093|issue:41762", + "issue:43296|issue:43366" + ], + "prepared_review_unit_hash": "afe20e379c359238cd511977591dd626b5ca5622b4c7d549505e2f64326a9b49", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 63, + "serialized_chars": 46482, + "estimated_input_tokens": 11621, + "estimated_eval_tokens": 23498 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous; only the DeepSpeed ZeRO-3 index-0 failures look like true duplicates. The rest are different model/config/documentation issues with similar wording but distinct root causes.", + "confidence": 0.79, + "canonical_issue_reason": "43638 is the best representative of the repeated DeepSpeed ZeRO-3 index-0-out-of-bounds bug; it most closely matches 43596 and is generic enough to cover the same failure mode.", + "canonical_pr_reason": null, + "best_issue_reason": "43638 has the clearest duplicate signal and the most reusable description for the shared ZeRO-3 index-0 bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different failures: Qwen3.5 config attribute error vs streaming chat/completions runtime failure." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Different models and symptoms; dtype mismatch on Mistral-3 preprocessing is not the same as GPT-OSS failing on AMD GPUs." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Both are model/runtime problems, but the autocast dtype mismatch and AMD GPU failure are distinct bugs." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Different code paths: AMD GPU runtime failure vs invalid method call in GraniteMoEHybrid." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Qwen-3 NVFP4 runtime error is unrelated to the GPT-OSS AMD GPU issue." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed collective mismatch and ONNX export error are unrelated failures." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ModernBERT ONNX export and Qwen3MoE FP8 failure are different model/export paths." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "ONNX export vs auto device mapping CUDA assert are not the same bug." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are v5 docs cleanup issues, but they cover different missing references and not one underlying defect." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "Export error and GenerationMixin import error are unrelated." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Similar index/shape wording, but one is ZeRO-3 index-0-out-of-bounds and the other is a mask-vs-tensor shape mismatch." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Different concrete errors: ZeRO-3 index-0-out-of-bounds vs mask shape mismatch." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Mask shape mismatch is unrelated to GPT-OSS tensor-parallel issues." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Tokenizer mapping failure is a different bug from the mask/index mismatch error." + }, + { + "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 one is a collective mismatch and the other is an index-0-out-of-bounds init error." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 index-0-out-of-bounds failure mode on Bert; very likely duplicate reports of the same bug." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "Detokenization bug and LoRA merge/reload inconsistency are different problems." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross_entropy being called is unrelated to GPT-OSS tensor-parallel loading." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Qwen-3 NVFP4 runtime error and LoRA save/reload mismatch are separate issues." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "Activation checkpointing/FSDP recomputation mismatch is not the same as the mask shape mismatch bug." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "Both mention DeepSpeed, but the training weight-dimension error and collective mismatch are different failure points." + }, + { + "left": "issue:43975", + "right": "issue:45237", + "accept": false, + "reason": "Different model/runtime paths; detokenization is unrelated to AMD GPU execution failures." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "ZeRO-3 Bert init bug is unrelated to GPT-OSS tensor-parallel support." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "Checkpointing tensor-size mismatch and custom trainer signature mismatch are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "Different bugs and different subsystems." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Mistral-3 dtype mismatch and Qwen3-vl video StopIteration are unrelated." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Gemma3 sliding_window_pattern and Mistral tokenizer regex patch are separate code paths." + }, + { + "left": "issue:43828", + "right": "issue:45161", + "accept": false, + "reason": "Autocast dtype mismatch is not the same as GPT-OSS TP failure." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch version incompatibility and Qwen-3 NVFP4 runtime error are unrelated." + }, + { + "left": "issue:44918", + "right": "issue:45161", + "accept": false, + "reason": "TRL SFT embedding unpacking failure is different from GPT-OSS tensor-parallel behavior." + }, + { + "left": "issue:34928", + "right": "issue:41720", + "accept": false, + "reason": "FSDP/activation checkpointing mismatch is unrelated to the CUDA assert on Qwen3 auto device mapping." + }, + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs issues about v5 removals, but they target different removed references." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "TextClassificationPipeline docs and run-command docs are distinct documentation problems." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Gemma3 config attribute error and Qwen3 CUDA assert are unrelated." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Corrupted Ovis2 checkpoint and NemotronH checkpoint loading failure are different issues." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Unit test loading failure and corrupted checkpoint are not the same bug." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Same model family, but different APIs and failures; not the same underlying bug." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper runtime failure and PretrainedConfig pydantic breakage are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Grouped_mm tracing failure and autocast dtype mismatch are different precision/tracing bugs." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "GenerationMixin import error and Gemma3 config attribute error are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure for Voxtral and compile error on non-template nodes are unrelated." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Mistral-3 preprocessing dtype mismatch and GraniteMoEHybrid invalid method are different bugs." + }, + { + "left": "issue:43828", + "right": "issue:44292", + "accept": false, + "reason": "Autocast dtype mismatch and Qwen-3 NVFP4 runtime error are distinct failures." + }, + { + "left": "issue:43039", + "right": "issue:44918", + "accept": false, + "reason": "Liger Kernel cross_entropy behavior and TRL SFT embedding unpacking are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": true, + "reason": "Both report the same ZeRO-3 index-0-out-of-bounds failure when loading/training with size-0 tensors; likely the same underlying bug." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs regressions, but they refer to different removed features and separate fixes." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Gemma3 sliding_window_pattern and Qwen3 auto device mapping CUDA assert are unrelated." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Checkpoint corruption and NemotronH load failure are different issues." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Test loading problem and corrupted checkpoint are not the same root cause." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Different Qwen3.5 scenarios and errors; not one duplicate bug." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper runtime failure and pydantic model breakage are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Different precision/tracing failures in different model paths." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "ImportError and config attribute error are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure and template compile error are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The duplicate calls on 43596/43638 and 41762/43638 are supported by the titles, and the rest of the pairwise rejections generally avoid overclaiming. I didn\u2019t see a soft-edge verdict that clearly overstates the evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 2, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36331", + "issue:39186", + "issue:39290", + "issue:39692", + "issue:40990", + "issue:41093", + "issue:41553", + "issue:41762", + "issue:42915", + "issue:43054", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43644", + "issue:43645", + "issue:43742", + "issue:43828", + "issue:43866", + "issue:43873", + "issue:43883", + "issue:43950", + "issue:43975", + "issue:43994", + "issue:44291", + "issue:44292", + "issue:44360", + "issue:44387", + "issue:44403", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44534", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44863", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:45005", + "issue:45070", + "issue:45161", + "issue:45399", + "issue:45464", + "issue:45507" + ], + "soft_pairs": [ + "issue:44387|issue:45005", + "issue:43572|issue:45070", + "issue:43541|issue:45161", + "issue:43828|issue:45464", + "issue:44291|issue:45070", + "issue:43366|issue:45464", + "issue:43531|issue:45070", + "issue:44661|issue:45464", + "issue:43054|issue:43994", + "issue:39290|issue:43742", + "issue:43541|issue:44589", + "issue:41093|issue:43596", + "issue:42915|issue:43541", + "issue:40990|issue:44960", + "issue:39186|issue:44918", + "issue:43404|issue:45005", + "issue:43645|issue:45399", + "issue:44530|issue:45005", + "issue:44863|issue:45005", + "issue:43873|issue:45005", + "issue:44360|issue:44512", + "issue:42915|issue:45399", + "issue:44918|issue:45507", + "issue:43975|issue:44292", + "issue:44492|issue:44512", + "issue:43950|issue:44534", + "issue:43644|issue:43950", + "issue:44292|issue:45507", + "issue:43828|issue:44928", + "issue:39692|issue:43054", + "issue:41553|issue:43883", + "issue:43866|issue:44403", + "issue:36331|issue:41762", + "issue:44560|issue:45507" + ], + "prepared_review_unit_hash": "b2e577c6f3d2140d6913596dd66ccb5e139223286084d688ac13ff9a801f8487", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37563, + "estimated_input_tokens": 9391, + "estimated_eval_tokens": 19038 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most items are unrelated model- or regression-specific issues bundled into one loose cluster. The only clear duplicate subcluster is the non-persistent buffer corruption regression in v5.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43950 is the most specific and diagnostic report of the non-persistent buffer regression, with an explicit v5.x corruption symptom and a clear `from_pretrained()` failure mode.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43950 best represents the only clearly cohesive duplicate set in this cluster; it is the most complete statement of the buffer-corruption regression.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Int4 quantization/OOM regression vs tied-weights translation-model issue; different failure modes." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Missing `pad_token_idx` in StableLmConfig is separate from the pydantic `PretrainedConfig` field break." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Torch dynamo grouped_mm tracing crash vs GPT-OSS tensor-parallel support problem; not the same bug." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Autocast dtype mismatch on Phi MoE vs Qwen3.5 streaming chat/completions failure." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "Unexpected `_is_hf_initialized` arg during `init_empty_weights` is distinct from the pydantic config regression." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "GGUF architecture support request for gpt-oss is not the same as a Qwen3.5 streaming inference bug." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Qwen3-MoE `sliding_window` bug is unrelated to `PretrainedConfig`/pydantic breakage." + }, + { + "left": "issue:44661", + "right": "issue:45464", + "accept": false, + "reason": "Tokenizer-mapping failure in `add-new-model-like` is a different code path from streaming inference." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": false, + "reason": "Both involve SigLIP2, but one reports degraded embeddings and the other nonsensical AutoModel/pipeline output; too little evidence of one underlying bug." + }, + { + "left": "issue:39290", + "right": "issue:43742", + "accept": false, + "reason": "Gemma3 `sliding_window_pattern` missing attribute vs MobileLLM key-error on load; different model/config issues." + }, + { + "left": "issue:43541", + "right": "issue:44589", + "accept": false, + "reason": "Grouped_mm tracing failure and missing Float8 storage are unrelated runtime errors." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Mask-shape mismatch in one path vs DeepSpeed ZeRO-3 zero-sized tensor indexing in another." + }, + { + "left": "issue:42915", + "right": "issue:43541", + "accept": false, + "reason": "FineGrainedFP8Config failure on Qwen3Moe is not the same as Mixtral grouped_mm tracing failure." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "Perplexity regression on gpt-oss-20b vs a GLM5 issue; no shared underlying bug shown." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP `'weight' must be 2-D` and TRL embedding unpacking failure are distinct training-time errors." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Both mention tied weights, but the affected models and reported failures differ too much to merge as one bug." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Notebook custom-model initialization breakage is unrelated to flash-attn2 fallback checks." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache group-type crash is not the same underlying issue as tied weights." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading failure is unrelated to translation-model tied weights." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization/offloading behavior and tied-weights regression are different problems." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "A DSA-indexer discussion/joke is not the same as docs mentioning a removed run command." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Qwen3Moe FP8 support failure vs flash-attn fallback gating; different fixes." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "TRL SFT embedding unpacking error and GraniteMoEHybrid invalid-method call are unrelated." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Detokenization bug in deepseek-coder is separate from the Qwen-3-8B NVFP4 runtime error." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "Cache-strategy typo and stale docs about a removed command are unrelated documentation issues." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Both report the same v5 regression where non-persistent buffers are corrupted/filled with junk after loading." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": true, + "reason": "Same non-persistent-buffer corruption regression; 43950 is the more specific description of the bug." + }, + { + "left": "issue:44292", + "right": "issue:45507", + "accept": false, + "reason": "NVFP4 runtime failure and GraniteMoEHybrid invalid-method error do not appear to share a code path." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Autocast dtype mismatch on Phi MoE vs RLHF NaN explosion from 3D position_ids and SDPA fallback." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "SigLIP2 docs example problems are different from a report about weak text embeddings." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Bad AutoTokenizer error messaging for Voxtral is separate from missing `all_tied_weights_keys` on MolmoForCausalLM." + }, + { + "left": "issue:43866", + "right": "issue:44403", + "accept": false, + "reason": "Corrupted Ovis2 checkpoint vs generic loading noise; not the same defect." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Trainer `compute_loss()` kwarg breakage is unrelated to DeepSpeed ZeRO-3 loading/index errors." + }, + { + "left": "issue:44560", + "right": "issue:45507", + "accept": false, + "reason": "Qwen3-vl video `StopIteration` and GraniteMoEHybrid invalid-method issues are distinct." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet and stays conservative. The only clearly supported duplicate subcluster is the non-persistent buffer corruption regression (issue:43644 / issue:43950 / issue:44534), and the rest of the pairwise rejections avoid over-merging unlike issues. No obvious evidence overstatement or unmergeable PR-pair claim." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 3, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36331", + "issue:38175", + "issue:41093", + "issue:41553", + "issue:41762", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43257", + "issue:43278", + "issue:43404", + "issue:43493", + "issue:43643", + "issue:43742", + "issue:43782", + "issue:43792", + "issue:43824", + "issue:43827", + "issue:43854", + "issue:43856", + "issue:43866", + "issue:43872", + "issue:43881", + "issue:43883", + "issue:43940", + "issue:43950", + "issue:43975", + "issue:44220", + "issue:44291", + "issue:44368", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44509", + "issue:44589", + "issue:44661", + "issue:44863", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45305", + "issue:45313", + "issue:45356", + "issue:45399", + "issue:45520" + ], + "soft_pairs": [ + "issue:43872|issue:44589", + "issue:41553|issue:43881", + "issue:38175|issue:43054", + "issue:42947|issue:44387", + "issue:43054|issue:43493", + "issue:43881|issue:43883", + "issue:44403|issue:44451", + "issue:42947|issue:43856", + "issue:43643|issue:45020", + "issue:43975|issue:45520", + "issue:43782|issue:44960", + "issue:44403|issue:44488", + "issue:42491|issue:44368", + "issue:41762|issue:42915", + "issue:43975|issue:44451", + "issue:45020|issue:45356", + "issue:44661|issue:45020", + "issue:43054|issue:43278", + "issue:43824|issue:43872", + "issue:44991|issue:45399", + "issue:43827|issue:44509", + "issue:43792|issue:44220", + "issue:45081|issue:45399", + "issue:43404|issue:43950", + "issue:43257|issue:43866", + "issue:42915|issue:45020", + "issue:43742|issue:45399", + "issue:43742|issue:45020", + "issue:41553|issue:44991", + "issue:43940|issue:45313", + "issue:43872|issue:44291", + "issue:43854|issue:44863", + "issue:43257|issue:45305", + "issue:41553|issue:44488", + "issue:42915|issue:43872", + "issue:36331|issue:41093" + ], + "prepared_review_unit_hash": "9955d509fcfbab0e0755a5f6c0f20634de886f6fd6badc5a5c545906cb771698", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 38581, + "estimated_input_tokens": 9646, + "estimated_eval_tokens": 19548 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of model-loading and v5 regression reports across unrelated models. Only a few soft edges look like true duplicates: the docs pipeline-removal reports, the SigLIP2 implementation discrepancy pair, and the Whisper fbank/extract feature path pair.", + "confidence": 0.69, + "canonical_issue_reason": "No single perfect canonical issue exists, but issue 45020 is the broadest umbrella in the set: recent-version / remote_code breakage during model loading. It is the best representative of the cluster\u2019s dominant theme, though many items are still distinct bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "45020 is the best global anchor because it captures the broadest and most central loading/regression pattern in the cluster and can contextually group several nearby reports, even if it does not subsume every issue here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Different bugs: bitsandbytes init-arg incompatibility vs missing Float8 storage registration." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Both are model-loading failures, but for different models and different failure modes." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Same SigLIP2 family, but zero probabilities and degraded text embeddings are different symptoms without enough evidence of one root cause." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Both involve memory, but one is gradient checkpointing/LoRA ineffective and the other is int4 quantization OOM." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": true, + "reason": "Same SigLIP2 text-embedding regression/discrepancy; the HF-vs-JAX mismatch plausibly explains the degraded embeddings." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "Unrelated failures: GLM-4V loading vs Molmo missing all_tied_weights_keys." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "Generic loading noise vs a specific model load failure; not the same underlying bug." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Different Qwen3 MoE training issues with different causes and symptoms." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "45020 is a broad umbrella; 43643 is a specific AutoConfig/trust_remote_code symptom, not clearly the same bug." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Different subsystems entirely: detokenization regression vs flash_attn import key error." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Different models and different load-time failures." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Qwen2.5-VL import error vs bitsandbytes _is_hf_initialized mismatch." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Both mention loading, but one is docs/UX noise while the other is a concrete model load failure." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "Both are Qwen LoRA-related, but one is cross-version model usability and the other is a tie_word_embeddings warning." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Different DeepSpeed/Qwen problems: ZeRO-3 load failure vs FineGrainedFP8Config failure." + }, + { + "left": "issue:43975", + "right": "issue:44451", + "accept": false, + "reason": "Different model-specific tokenizer/detokenization issues." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Tokenizer codec/warning regression is not the same as broad remote_code breakage." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Model-class mapping failure vs broad remote_code breakage; related area, not same bug." + }, + { + "left": "issue:43054", + "right": "issue:43278", + "accept": false, + "reason": "Both touch embeddings, but one is quality regression and the other is dtype drift between train/eval." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Same as above: import error vs bitsandbytes constructor mismatch." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading regression vs flash-attn fallback gating; different code paths." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Same docs regression: both still reference pipeline tasks removed in v5." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": true, + "reason": "Both point to the Whisper audio feature-extraction/fbank path failing." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer regex crash vs flash-attn fallback checks; unrelated failures." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Different bugs: untied lm_head weights vs non-persistent buffer corruption in from_pretrained." + }, + { + "left": "issue:43257", + "right": "issue:43866", + "accept": false, + "reason": "Different models and unrelated loading/corruption reports." + }, + { + "left": "issue:42915", + "right": "issue:45020", + "accept": false, + "reason": "Qwen3 MoE FP8 training issue vs broad remote_code breakage; not the same underlying bug." + }, + { + "left": "issue:43742", + "right": "issue:45399", + "accept": false, + "reason": "MobileLLM key error vs flash-attn fallback blocking; different subsystems." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "Specific model key error is not the same as broad remote_code regression." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Bad error message for one model vs tokenizer load failure for another; not duplicate." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": false, + "reason": "Both are ZeRO-3 weight-loading failures, but different Qwen variants and likely different root causes." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": false, + "reason": "Different bitsandbytes/init-empty-weights incompatibilities with different error signatures." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-loading failures for different models." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Qwen3 MoE accelerate/deepspeed weight conversion vs DeepSpeed gradient averaging; related area, not same bug." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Different models and different failure types." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Trainer compute_loss API mismatch vs tensor mask shape mismatch; unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative about the cluster being a mix of unrelated loading/regression issues. The only accepted soft edges are well-supported by the titles (docs pipeline removals, SigLIP2 discrepancy, Whisper fbank path), and the rejected pairs are generally framed cautiously rather than overstated. 45020 is a reasonable broad anchor, with appropriate caveats that it does not subsume the whole cluster." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 4, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36010", + "issue:38175", + "issue:39692", + "issue:41093", + "issue:42491", + "issue:42617", + "issue:42915", + "issue:43317", + "issue:43366", + "issue:43531", + "issue:43550", + "issue:43596", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43720", + "issue:43824", + "issue:43828", + "issue:43872", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44075", + "issue:44451", + "issue:44479", + "issue:44488", + "issue:44530", + "issue:44534", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44910", + "issue:44918", + "issue:44960", + "issue:44977", + "issue:45030", + "issue:45313", + "issue:45325", + "issue:45357", + "issue:45538" + ], + "soft_pairs": [ + "issue:41093|issue:44805", + "issue:44479|issue:44560", + "issue:43872|issue:45538", + "issue:44530|issue:44863", + "issue:43366|issue:43828", + "issue:24643|issue:30064", + "issue:43366|issue:43531", + "issue:43550|issue:44910", + "issue:43645|issue:43646", + "issue:43950|issue:45325", + "issue:43366|issue:44589", + "issue:44560|issue:44918", + "issue:43824|issue:43957", + "issue:42915|issue:43957", + "issue:42617|issue:43366", + "issue:38175|issue:43994", + "issue:43646|issue:43950", + "issue:44560|issue:44805", + "issue:36010|issue:42915", + "issue:43638|issue:44661", + "issue:44877|issue:45030", + "issue:43596|issue:44661", + "issue:43644|issue:44534", + "issue:43317|issue:44756", + "issue:45313|issue:45357", + "issue:38175|issue:39692", + "issue:42491|issue:43720", + "issue:44960|issue:44977", + "issue:44451|issue:44488", + "issue:43708|issue:44075" + ], + "prepared_review_unit_hash": "31e1998fd2fad8db2d1200aec73d9372eb673c7520456010259c4dbe642fad94", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36648, + "estimated_input_tokens": 9162, + "estimated_eval_tokens": 18580 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous issue set. I\u2019d keep only two soft duplicate pairs as accepted: the mask-shape mismatch reports and the non-persistent-buffer corruption reports. The other pairs are related by model family or broad subsystem in some cases, but the evidence here is not strong enough to treat them as the same underlying bug.", + "confidence": 0.77, + "canonical_issue_reason": "Issue 43644 is the clearest canonical report for the non-persistent-buffer regression: it describes the Transformers v5 junk-fill behavior directly, and 44534 appears to be the same bug with a near-identical title.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43644 is the strongest representative issue in this set because it is explicit, actionable, and has a close duplicate in 44534; it best anchors the non-persistent-buffer regression cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "Same mask/index shape-mismatch failure; 44805 is the broader wording of the same IndexError." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": false, + "reason": "Both involve Qwen video paths, but the titles point to different regressions and not one concrete shared bug." + }, + { + "left": "issue:43872", + "right": "issue:45538", + "accept": false, + "reason": "Different subsystems and failures: bitsandbytes storage/init incompatibility vs tokenizer max-length behavior." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "Unrelated problems: PagedAttentionCache/Qwen3.5 crash vs NemotronH checkpoint loading." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "Different bugs: GGUF/gpt-oss support vs Phi-tiny-MoE dtype mismatch under autocast." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "No shared code path: DeepSpeed weight-shape error vs void segmentation-map processing." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "Only a loose Qwen3-MoE association; one is GGUF support, the other is a sliding-window bug." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different model families and mechanisms: Bamba torch.compile/SDPA vs Qwen3.5 flash-attention position_ids handling." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": false, + "reason": "Related Transformers 5.0 custom-model breakage, but the reported failure modes are not specific enough to merge as one bug." + }, + { + "left": "issue:43950", + "right": "issue:45325", + "accept": false, + "reason": "Different regressions: non-persistent-buffer corruption vs Qwen2.5-VL rope/index scaling." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "Broad model-loading theme only; gpt-oss GGUF support is unrelated to missing Float8 storage." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Different failures in different paths: video StopIteration vs embedding unpacking during TRL SFT training." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Both are load/import regressions, but they affect different classes of models and code paths." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Qwen3Moe FP8 failure and meta-device loading issues are distinct defects." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "3d_parallel.py runtime issue is unrelated to GGUF gpt-oss support." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": false, + "reason": "Same model family and symptom class, but the evidence only shows similar bad outputs, not the same concrete bug." + }, + { + "left": "issue:43646", + "right": "issue:43950", + "accept": false, + "reason": "Custom model initialization breakage and non-persistent-buffer corruption are different regressions." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Video StopIteration and mask/tensor shape mismatch are unrelated failures." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "ImportError for GenerationMixin is unrelated to Qwen3Moe FP8 config failure." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Different code paths: DeepSpeed ZeRO-3 empty-index failure vs tokenizer-mapping-name handling." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both are config-loading regressions, but they target different models and validation failures." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "DeepSpeed ZeRO-3 BertModel failure is unrelated to add-new-model-like/tokenizer mapping." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Near-identical reports of the same Transformers v5 non-persistent-buffer junk-fill regression." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "Different loading/memory issues: dequantized offload loading vs Strix Halo mmap/OOM." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Separate Qwen3.5 regressions: ZeRO-3 weight loading vs incorrect save_pretrained visual-encoder keys." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Both involve SigLIP2, but one is output quality and the other is a docs/example mismatch plus quantization failure." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Different model-loading regressions: Qwen3-MoE LoRA compatibility vs BitNet packed-weight unpacking." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated models and failures: GLM5 vs Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-load failures, but they point to different checkpoints and no shared defect is established." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer resume/max_steps accounting bug is unrelated to SGD argument handling." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The writeup is mostly grounded, but it overstates duplicate certainty in a couple of soft-edge calls. In particular, issue:44479 vs issue:44560 and issue:38175 vs issue:43994 are only loosely supported by the titles alone; they look related by model family / symptom class, but not clearly the same bug. The summary also phrases these as \"clear duplicate pairs,\" which is stronger than the packet evidence justifies. The safer accepted pairs are issue:41093 vs issue:44805 and issue:43644 vs issue:44534; the rest should stay conservative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 5, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43381", + "issue:43404", + "issue:43425", + "issue:43454", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43828", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44292", + "issue:44387", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44488", + "issue:44661", + "issue:44683", + "issue:44743", + "issue:44756", + "issue:44863", + "issue:44877", + "issue:44912", + "issue:44928", + "issue:44960", + "issue:44991", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45362", + "issue:45406" + ], + "soft_pairs": [ + "issue:43854|issue:43931", + "issue:43856|issue:43873", + "issue:42491|issue:45276", + "issue:41720|issue:42915", + "issue:43257|issue:43854", + "issue:39692|issue:43994", + "issue:43856|issue:44387", + "issue:44488|issue:44991", + "issue:45362|issue:45406", + "issue:43644|issue:43645", + "issue:42915|issue:44756", + "issue:42491|issue:43931", + "issue:44661|issue:44683", + "issue:45276|issue:45335", + "issue:42491|issue:43856", + "issue:43404|issue:43454", + "issue:36010|issue:39404", + "issue:44164|issue:44291", + "issue:41720|issue:44155", + "issue:44402|issue:44488", + "issue:43873|issue:44387", + "issue:39692|issue:44960", + "issue:44387|issue:44756", + "issue:43425|issue:43957", + "issue:44292|issue:44912", + "issue:42947|issue:43381", + "issue:43854|issue:45216", + "issue:44291|issue:44756", + "issue:45276|issue:45292", + "issue:44743|issue:44960", + "issue:43381|issue:44928", + "issue:43957|issue:44756", + "issue:44164|issue:44756", + "issue:43646|issue:43708", + "issue:45292|issue:45335", + "issue:44479|issue:44877", + "issue:44402|issue:44451", + "issue:43645|issue:44164", + "issue:43257|issue:44863", + "issue:43425|issue:43828", + "issue:43828|issue:43957" + ], + "prepared_review_unit_hash": "1a6e86bc21033eaca11499ada59c91691e59c965eadb7f5b81a8eb94ddf772af", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 41, + "serialized_chars": 39812, + "estimated_input_tokens": 9953, + "estimated_eval_tokens": 20162 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The batch contains multiple loose groups, with at least two coherent duplicate families: a resize_token_embeddings propagation family (45292/45276/45335) and a separate lm_head weight-tying family (43404/43454). 45292 is the best representative for the resize_token_embeddings family, but the cluster is not single-family.", + "confidence": 0.79, + "canonical_issue_reason": "issue:45292 is the best representative of the resize_token_embeddings family because it describes the generic propagation failure most clearly, with 45276 and 45335 as close model-specific variants.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45292 is the strongest issue candidate for one coherent duplicate family, but not the only one; the batch also contains a plausible separate lm_head weight-tying family around 43404/43454.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Different model-loading failures; same broad area, not the same bug." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Both involve memory/quantization, but the symptoms and code paths differ." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "LoRA checkpoint compatibility vs resize_token_embeddings propagation are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Different Qwen3 failure modes: cuda assert vs FP8 config loading." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Separate model-loading bugs for different models and mechanisms." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Same model family, but one is a docs/example issue and the other is an inference correctness problem." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Both mention memory pressure, but they are distinct regressions." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-load failures, but for different models with no clear shared concrete bug." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Completely different components: chat crash vs processor attribute error." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Different Transformers 5.0 regressions with different failure surfaces." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "FP8 loading failure vs mmap/OOM workaround are unrelated." + }, + { + "left": "issue:42491", + "right": "issue:43931", + "accept": false, + "reason": "Different Qwen3 checkpoint/model-loading problems." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Tokenizer mapping bug vs torch flex_attention compile failure are unrelated." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Same resize_token_embeddings regression: resized embeddings are not propagated to model-specific output/decoder embeddings." + }, + { + "left": "issue:42491", + "right": "issue:43856", + "accept": false, + "reason": "Different Qwen3 MoE checkpoint compatibility and memory-usage problems." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": true, + "reason": "Same lm_head weight-tying defect showing up in two model classes; one fix could plausibly cover both." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "ImportError for GenerationMixin and Whisper pipeline return_language are unrelated." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different serialization/init regressions: extra_state handling vs init_empty_weights argument mismatch." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different models and failures; no shared code-path evidence." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "Both are loading/tokenizer-related, but the reported model-specific failures differ." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Quantization/offloading memory issues, but not the same concrete defect." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "SigLIP2 example bug and GLM5 are unrelated." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Both are OOM-related, but the triggers and fixes are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch version compatibility and meta-device loading are distinct problems." + }, + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Different quantization formats and model families." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Different gradient-checkpointing issues with different conditions." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "Different models and different failure modes: loading vs save_pretrained regression." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "Unexpected init argument vs Strix Halo mmap/OOM are unrelated." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same resize_token_embeddings family; 45292 is the broad generic report and 45276 is a model-specific manifestation." + }, + { + "left": "issue:44743", + "right": "issue:44960", + "accept": false, + "reason": "Recurrent-state cache bug and GLM5 are unrelated." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Gradient-checkpointing eval-mode restriction vs RLHF NaN explosion are different bugs." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device loading failure vs mmap/OOM workaround are unrelated." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "Serialization extra_state bug vs Strix Halo memory issue are unrelated." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization regression and resume_from_checkpoint max_steps bug are unrelated." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Same resize_token_embeddings propagation bug affecting output/decoder embeddings." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Video-input regression and strict config loading are unrelated." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Tokenizer vocab mismatch and model load failure are not clearly the same defect." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Notebook/custom-model init issue and extra_state save/load issue are separate regressions." + }, + { + "left": "issue:43257", + "right": "issue:44863", + "accept": false, + "reason": "Qwen3 MoE deepspeed loading and NemotronH checkpoint loading are different model-specific bugs." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch compatibility issue and autocast dtype mismatch are unrelated." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Autocast dtype mismatch and meta-device loading failure are different bugs." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "Reject: the packet supports at least two coherent duplicate families, not just one. The resize_token_embeddings set is grounded, but the best_issue_reason overstates uniqueness by calling it the only strongly coherent duplicate set. Also, the lm_head weight-tying pair (43404/43454) is itself a plausible duplicate family, so the summary should be more conservative about cluster structure." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 6, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:32090", + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42947", + "issue:43299", + "issue:43421", + "issue:43425", + "issue:43452", + "issue:43582", + "issue:43596", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43856", + "issue:43883", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44206", + "issue:44220", + "issue:44265", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44671", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44898", + "issue:44912", + "issue:44918", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45161" + ], + "soft_pairs": [ + "issue:44589|issue:44756", + "issue:43906|issue:44530", + "issue:43931|issue:44863", + "issue:39404|issue:44912", + "issue:43611|issue:43645", + "issue:44530|issue:45084", + "issue:43425|issue:43716", + "issue:44265|issue:44898", + "issue:43901|issue:44509", + "issue:43299|issue:43931", + "issue:44479|issue:44805", + "issue:43994|issue:44079", + "issue:44671|issue:44991", + "issue:43856|issue:44756", + "issue:39404|issue:44671", + "issue:40990|issue:42947", + "issue:44589|issue:45084", + "issue:43596|issue:44683", + "issue:43425|issue:44918", + "issue:36032|issue:43452", + "issue:32090|issue:39404", + "issue:43582|issue:43632", + "issue:44991|issue:45081", + "issue:40990|issue:43421", + "issue:43957|issue:44164", + "issue:44164|issue:44479", + "issue:44488|issue:45081", + "issue:43883|issue:44488", + "issue:43425|issue:45161", + "issue:43611|issue:43644", + "issue:44206|issue:44220", + "issue:42947|issue:44756", + "issue:44488|issue:44987", + "issue:34567|issue:40444", + "issue:43716|issue:45084" + ], + "prepared_review_unit_hash": "d315a8e1dee27dd5881a7440280ae534ce3c476401cce9376c2704d404937530", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38181, + "estimated_input_tokens": 9546, + "estimated_eval_tokens": 19348 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a heterogeneous grab-bag of Transformers v5 regression reports, with many unrelated model/tokenizer/training issues. I do not see any soft-edge pair that looks like a true duplicate, so all soft edges are rejected.", + "confidence": 0.72, + "canonical_issue_reason": "issue:43611 is the broadest, most general representative here: a Transformers v5 model-loading regression. It best fits the recurring pattern across the cluster better than the narrower model- or tokenizer-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43611 is the best umbrella issue for this cluster because it describes the most general load-path regression and can absorb the broad duplicate chatter around v5 loading failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Different failure classes: storage lookup/type error vs mmap-related OOM avoidance. No shared code path." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "An isolated reproduction of a prior issue is not the same as a PagedAttentionCache linear_attention crash; different models and code paths." + }, + { + "left": "issue:43931", + "right": "issue:44863", + "accept": false, + "reason": "Both are model-loading complaints, but they target different model families and failure modes (shape mismatch vs checkpoint loading support)." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline language-return regression is unrelated to MXFP4 quantization fallback/loading behavior." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both involve v5 loading regressions, but one is base_model_prefix/model loading and the other is custom model initialization in notebooks; not the same bug." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3.5 cache crash and template-node compilation error are unrelated issues." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Torch version incompatibility and an image preprocessor dtype mismatch are separate problems." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export with torch_compilable_check is a different failure path from Perceiver interpolation handling." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs issues, but they cover different documentation gaps and different pipeline/task content." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Same broad area (Qwen/VL loading) but different concrete bugs: MoE loading breakage vs checkpoint shape mismatch." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Video-input regression and mask-shape IndexError are not the same underlying bug." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "One is a bad-results complaint for a specific model; the other is a ModelOutput key assignment bug." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT prediction correctness and tokenizer loading for est-roberta are unrelated." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "MoE training memory inefficiency is unrelated to disabling mmap on Strix Halo to avoid OOM." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline language output and CamemBERT masked-LM predictions are different model-specific regressions." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "Perplexity anomaly and gradient-checkpointing ineffectiveness are distinct training/optimization issues." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Float8 storage lookup error and template-node compilation failure are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "DeepSpeed ZeRO-3 initialization index error and compiled flex_attention on torch>=2.9 are different code paths." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch 2.10 incompatibility and TRL input-embedding unpacking failure are unrelated regressions." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "T5 tokenizer special-token conflict and gguf_file/from_pretrained breakage are not the same bug." + }, + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error and Whisper return_language pipeline regression are unrelated." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Apple Silicon warmup TypeError and _is_hf_initialized flag breakage are different issues." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Both are tokenizer-loading problems, but one is a specific tokenizer load failure while the other is a Mistral regex patch crash with a different trigger." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Perplexity regression and runtime special-token post-processor updates are unrelated." + }, + { + "left": "issue:43957", + "right": "issue:44164", + "accept": false, + "reason": "Meta-device loading breakage and extra_state save/from_pretrained handling are separate model lifecycle bugs." + }, + { + "left": "issue:44164", + "right": "issue:44479", + "accept": false, + "reason": "Extra_state serialization and video-input regression have no common concrete failure path." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer load failure for a specific model and Mistral regex patch crash are different issues." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing all_tied_weights_keys on Molmo and tokenizer-loading failure for sleng-bert are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:45161", + "accept": false, + "reason": "Torch version incompatibility and GPT-OSS MoE tensor-parallel behavior are not the same bug." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "Both are v5 regressions, but one is model-loading/base_model_prefix and the other is non-persistent buffer initialization; different concrete problems." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Audio feature-extractor center-arg crash and _torch_extract_fbank_features() issue are related area-wise, but not clearly the same underlying defect." + }, + { + "left": "issue:42947", + "right": "issue:44756", + "accept": false, + "reason": "LoRA gradient-checkpointing behavior and Strix Halo mmap OOM are unrelated." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Loading sleng-bert and loading physical-intelligence/fast on newer transformers are different tokenizer/model-loading failures." + } + ] + }, + "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": 13, + "review_unit_index": 7, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36032", + "issue:39404", + "issue:40990", + "issue:41553", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43452", + "issue:43482", + "issue:43531", + "issue:43540", + "issue:43541", + "issue:43582", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43819", + "issue:43825", + "issue:43883", + "issue:43994", + "issue:44220", + "issue:44265", + "issue:44403", + "issue:44451", + "issue:44466", + "issue:44479", + "issue:44530", + "issue:44534", + "issue:44610", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44849", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45127", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45507" + ], + "soft_pairs": [ + "issue:43644|issue:43646", + "issue:43883|issue:44991", + "issue:43643|issue:44530", + "issue:43452|issue:43482", + "issue:36032|issue:42915", + "issue:43317|issue:43482", + "issue:43262|issue:43377", + "issue:44451|issue:44987", + "issue:43541|issue:45084", + "issue:39404|issue:43825", + "issue:42915|issue:44530", + "issue:44849|issue:45325", + "issue:43404|issue:44466", + "issue:36032|issue:43645", + "issue:43646|issue:44534", + "issue:44987|issue:44991", + "issue:45084|issue:45507", + "issue:43994|issue:44610", + "issue:43377|issue:43819", + "issue:43645|issue:44534", + "issue:45216|issue:45357", + "issue:43322|issue:43540", + "issue:44403|issue:45081", + "issue:44451|issue:45081", + "issue:44265|issue:44610", + "issue:43632|issue:44756", + "issue:41553|issue:43582", + "issue:44220|issue:44683", + "issue:43404|issue:45127", + "issue:42915|issue:45084", + "issue:44479|issue:45325", + "issue:44987|issue:45081", + "issue:43296|issue:43531", + "issue:40990|issue:44811" + ], + "prepared_review_unit_hash": "15f4b920270d163dc4a5f4a0c6aac4791d3343a0d6bd795100e2495900d44ee6", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37892, + "estimated_input_tokens": 9473, + "estimated_eval_tokens": 19202 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is too heterogeneous to treat as one duplicate set. The GGUF-loading and Qwen3.5 save_pretrained pairs look like real duplicates, but the Qwen video/rope pair is only loosely related and should not be accepted; overall the cluster should be rejected.", + "confidence": 0.94, + "canonical_issue_reason": "45216 is the clearest representative of the strongest subcluster: a precise Qwen3.5 save_pretrained regression with a direct follow-up in 45357.", + "canonical_pr_reason": null, + "best_issue_reason": "45216 is the best single issue in the packet because it is concrete, well-scoped, and has the strongest duplicate linkage.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Both are Transformers v5 regressions, but the bugs are different: non-persistent buffers vs custom model initialization." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Both are loading failures, but they affect different models and different failure modes." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "AutoConfig missing fields and a PagedAttentionCache crash are unrelated code paths." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": true, + "reason": "Both are GGUF-loading regressions; 43482 looks like a model-specific instance of the broader `gguf_file` breakage." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "Tokenizer method conflict and Qwen3Moe FP8 failure are unrelated." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "Device-map/offload dequantization failure is separate from GGUF loading." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and MIMI padding-mask batching are different bugs." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Similar loading complaints, but no evidence of the same root cause." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Torch dynamo grouped_mm tracing error and template compilation TypeError are different failures." + }, + { + "left": "issue:39404", + "right": "issue:43825", + "accept": false, + "reason": "Whisper return_language behavior and a pipeline error-message issue are distinct." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and Qwen3.5 cache/group-type crash are different problems." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen VL bugs: output_hidden_states behavior vs rope-index scaling for still-image temporal positions." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "Both involve tied weights, but one is a missing tie and the other is a save/serialization inconsistency." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer special-token conflict and notebook custom-model initialization breakage are unrelated." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Custom model initialization and junk non-persistent buffers are separate regressions." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Different affected models and no clear shared bug." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template-node TypeError vs GraniteMoEHybrid invalid method call are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "Nonsensical outputs and processor image-size mismatch are different issues." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "MIMI batching discrepancy and DAC from_latents mismatch are unrelated." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Notebook initialization breakage and buffer initialization junk are different bugs." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Same Qwen3.5 save_pretrained regression; 45357 is a later symptom focused on incorrect visual encoder keys." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Llava Next segmentation fault and Qwen3OmniMoe video ValueError are unrelated." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Loading noise and Mistral regex patch crash are different issues." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "ScandiBERT load failure and Mistral tokenizer regex crash are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export failure and OmDet-Turbo processor size mismatch are not the same bug." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "_is_hf_initialized flag regression and mmap/OOM on Strix Halo are unrelated." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Voxtral tokenizer error message and Apple Silicon allocator TypeError are unrelated." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Audio feature extraction and compiled flex_attention torch 2.9 failure are different problems." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Both touch tied weights, but one is a missing tie and the other is a LoRA-merge collapse case." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3Moe FP8 failure and template compilation TypeError are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "This is only loosely related by Qwen VL/video-positioning terminology; one is a broad video-input regression, the other a specific still-image temporal position_ids scaling bug." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Different model-loading/tokenizer problems with no evidence of the same root cause." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window issue are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "Perplexity anomaly and Whisper batch_decode skip_special_tokens handling are unrelated." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary is mostly grounded, but one accepted soft-edge verdict is too strong for the evidence in the packet. In particular, `issue:44479` vs `issue:45325` is only loosely related by Qwen VL/video-positioning terminology: one is a broad video-input regression across several Qwen models, while the other is a specific still-image temporal position_ids scaling bug. From titles alone, that should be treated as at most a plausible relation, not a true duplicate. The GGUF and Qwen3.5 save_pretrained pairs look more defensible, but this overreach means the cluster should be rejected." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 8, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:36032", + "issue:40444", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43317", + "issue:43377", + "issue:43421", + "issue:43425", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43646", + "issue:43742", + "issue:43749", + "issue:43792", + "issue:43818", + "issue:43856", + "issue:43881", + "issue:44062", + "issue:44079", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44488", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44683", + "issue:44829", + "issue:44877", + "issue:44987", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45491", + "issue:45538", + "issue:45542" + ], + "soft_pairs": [ + "issue:42175|issue:45542", + "issue:43611|issue:44534", + "issue:44062|issue:44987", + "issue:42915|issue:44291", + "issue:43317|issue:43856", + "issue:36032|issue:43742", + "issue:40444|issue:45325", + "issue:43632|issue:44534", + "issue:43531|issue:44877", + "issue:42915|issue:44589", + "issue:43421|issue:44568", + "issue:43818|issue:45357", + "issue:36032|issue:41720", + "issue:43482|issue:45538", + "issue:44829|issue:45479", + "issue:44683|issue:45020", + "issue:33666|issue:40444", + "issue:43425|issue:44403", + "issue:44403|issue:44991", + "issue:44079|issue:44534", + "issue:35707|issue:40444", + "issue:43377|issue:45491", + "issue:43792|issue:44589", + "issue:43482|issue:43525", + "issue:44291|issue:44589", + "issue:43749|issue:44079", + "issue:43644|issue:44079", + "issue:44079|issue:45072", + "issue:44265|issue:45072", + "issue:43632|issue:45072", + "issue:40444|issue:45381", + "issue:45325|issue:45381", + "issue:43611|issue:43646", + "issue:43531|issue:44291", + "issue:41720|issue:45081", + "issue:43881|issue:44488" + ], + "prepared_review_unit_hash": "19e64b88efc6a87d379bc22ec4008054f502a5ddc56839e8fe883f51e0215850", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 38422, + "estimated_input_tokens": 9606, + "estimated_eval_tokens": 19468 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "All soft pairs are false duplicates: they mostly share only a broad subsystem, model family, or Transformers-v5 regression theme, but each points to a different concrete failure mode. No single canonical issue stands out, and there is no PR evidence in this cluster.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue: the set spans unrelated bugs across tokenizers, loading, vision/video, device mapping, quantization, and training; the similarities are too broad to unify into one underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "No clear global best issue. A broad v5-regression report might look umbrella-like, but none of the issues cleanly subsumes the others as the same bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both concern TensorFlow-related install/backend behavior, but one is about missing TF in `pip install .[torch]` and the other about an undefined TF backend when only tensorboard is installed. Different packaging/config bugs." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "Both are Transformers v5 regressions, but one breaks loading models with `base_model_prefix` and the other corrupts non-persistent buffers. Different code paths and symptoms." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "The first is a tokenizer `AddedToken(..., special=...)` argument collision; the second is loading a specific model (`physical-intelligence/fast`) on newer Transformers. Too little evidence of the same defect." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "One is a FineGrainedFP8Config failure on Qwen3MoE; the other is an `_is_hf_initialized` argument error with `init_empty_weights`. Different initialization/quantization bugs." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "Both mention memory/offload concerns, but one is device_map dequantized loading on gpu+cpu offload and the other is inefficient Qwen3 MoE training memory use. Not the same bug." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "Tokenizer `add_special_tokens` conflict in T5Tokenizer versus a key error loading MobileLLM-125M. Same area only at a very high level." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "Both involve Qwen2.5-VL, but one is iterable-dataset training with multiple images per prompt and the other is incorrect still-image temporal position scaling in `get_rope_index`. Different failures." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "Both are v5 regressions, but `_is_hf_initialized` handling and junk-filled non-persistent buffers are separate loading/initialization issues." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "One is a Qwen3-MoE sliding-window issue; the other is strict config blocking `granite_speech`. No concrete overlap." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "FineGrainedFP8Config failure is unrelated to the `Float8_e4m3fnStorage` lookup/type error. Same quantization theme, different bug." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime post-processor updates for special tokens is not the same as `add_special_tokens=True` failing to add BOS/EOS for a specific tokenizer." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Video-LLaVA missing temporal attention/weight sharing is unrelated to Qwen3.5 `save_pretrained` writing wrong visual encoder keys." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Tokenizer load conflict and Qwen3 auto device mapping CUDA assert are unrelated failures." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Qwen2.5-GGUF loading with Transformers v5 and CLIPTokenizer `model_max_length` being 10**30 are different tokenizer/model-loading problems." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both concern degenerate classification training, but one is tied to `flash_attention_3` and the other to `num_labels=1`/`problem_type`. Not the same concrete issue." + }, + { + "left": "issue:44683", + "right": "issue:45020", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 and remote_code loading regressions are unrelated." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both are Qwen VL training issues, but multi-GPU training and iterable-dataset multi-image failures are distinct problems." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch 2.10 compatibility and generic loading noise are not the same bug." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Unnecessary loading noise and tokenizer load failure for EMBEDDIA/est-roberta are unrelated." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment with `None` versus non-persistent buffer corruption are separate internals issues." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is unrelated to Qwen2.5-VL iterable-dataset multi-image training." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Both involve batched vs single-input discrepancies, but MIMI encoder padding-mask support and Gemma3 sliding-window all-padding NaNs are different models and mechanisms." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper-large-v2 not running and `Float8_e4m3fnStorage` lookup failure are unrelated." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "GGUF loading under Transformers v5 and missing `pad_token_id` on `Llama4Config` are distinct loading/config issues." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "Unexpected `_is_hf_initialized` in init-empty-weights and missing float8 storage are separate v5-related failures." + }, + { + "left": "issue:43749", + "right": "issue:44079", + "accept": false, + "reason": "Broken FSDP CPU RAM efficient loading and misassigned ModelOutput keys are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer junk and ModelOutput key assignment bugs affect different code paths." + }, + { + "left": "issue:44079", + "right": "issue:45072", + "accept": false, + "reason": "ModelOutput key handling and dtype mismatches in SwitchTransformers/TimmWrapperModel are different issues." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "`torch.export.export` failure via `torch_compilable_check` is unrelated to bfloat16 dtype mismatches." + }, + { + "left": "issue:43632", + "right": "issue:45072", + "accept": false, + "reason": "`_is_hf_initialized` breakage and dtype mismatch CI failures are not the same bug." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL vision issues, but one is multi-image iterable training failure and the other is wrong video vision_position_ids. Different concrete paths." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Both concern Qwen2.5-VL position IDs, but one is still-image temporal scaling and the other is video input position IDs. Related area, not the same defect." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "`base_model_prefix` loading regression and custom model initialization breakage are separate Transformers v5 issues." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior and `_is_hf_initialized` init-empty-weights errors do not share the same code path." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert and Mistral tokenizer regex patch crash are unrelated." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Loading `glm-4v-9b` and loading `cjvt/sleng-bert` fail for different models and likely different causes." + } + ] + }, + "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": 13, + "review_unit_index": 9, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:39692", + "issue:40444", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43454", + "issue:43482", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43632", + "issue:43701", + "issue:43720", + "issue:43746", + "issue:43749", + "issue:43792", + "issue:43819", + "issue:43828", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:43957", + "issue:43994", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44387", + "issue:44466", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44683", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45325", + "issue:45357", + "issue:45563" + ], + "soft_pairs": [ + "issue:43454|issue:44466", + "issue:42175|issue:43296", + "issue:43994|issue:44898", + "issue:44387|issue:44617", + "issue:43828|issue:44683", + "issue:42915|issue:43482", + "issue:43526|issue:43883", + "issue:43701|issue:45325", + "issue:43262|issue:43746", + "issue:43792|issue:43957", + "issue:43632|issue:45563", + "issue:43317|issue:43873", + "issue:43606|issue:44265", + "issue:43881|issue:44991", + "issue:34634|issue:35707", + "issue:44186|issue:44265", + "issue:44291|issue:44877", + "issue:43606|issue:43819", + "issue:43377|issue:44898", + "issue:35707|issue:41720", + "issue:44610|issue:45325", + "issue:43454|issue:45072", + "issue:43454|issue:43550", + "issue:43377|issue:44610", + "issue:39692|issue:40444", + "issue:43749|issue:44534", + "issue:44466|issue:45357", + "issue:43296|issue:44991", + "issue:43454|issue:43720", + "issue:43296|issue:44877", + "issue:41720|issue:43322", + "issue:41720|issue:44977", + "issue:43792|issue:44683", + "issue:43404|issue:45357", + "issue:44186|issue:45072" + ], + "prepared_review_unit_hash": "857c036006fe2bc187070b5914c40e77798206c62427673b3c6d7a846a03fe01", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38166, + "estimated_input_tokens": 9542, + "estimated_eval_tokens": 19340 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated model-loading, quantization, and generation bugs. I found no true duplicate pairs; the soft matches are only broad subsystem similarity, not the same concrete bug or change.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43873 is the broadest still-open report here and best captures the shared offloading/quantization theme, but the cluster is too heterogeneous for a strong duplicate canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43873 is the best global anchor because it is broad, active, and has the clearest cross-cutting loading/offload scope among the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Both mention tied lm_head weights, but one is a model-specific tie bug and the other is a device-dependent serialization inconsistency; not the same concrete failure." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "Completely different problems: missing TensorFlow backend in an install extra vs a PaddleOCR-VL loading failure in vLLM." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Both are vision-input mismatches, but they involve different models and different root causes; too broad to treat as one bug." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Both end in OOM, but one is a Transformers int4 memory regression and the other is Sam3Video model memory usage; not the same issue." + }, + { + "left": "issue:43828", + "right": "issue:44683", + "accept": false, + "reason": "Dtype-mismatch under autocast and compiled flex_attention failure on new torch versions are distinct code paths." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Different models and different breakages: FP8 config failure vs GGUF loading failure under v5." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "One is a BeitImageProcessorFast label-reduction bug; the other is a missing all_tied_weights_keys attribute on MolmoForCausalLM." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint key mismatch and Qwen2.5-VL rope-index scaling are unrelated generation/loading bugs." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and GraniteSpeech PEFT checkpoint loading are different subsystems and bugs." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Whisper runtime failure vs meta-device loading regressions in v5 are separate model-loading problems." + }, + { + "left": "issue:43632", + "right": "issue:45563", + "accept": false, + "reason": "_is_hf_initialized flag breakage and a stale warning in paged generate are unrelated." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Both involve offloading, but one is a dequantized model load failure with device_map=auto and the other is a broader quantization/offload behavior report." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch and torch.export failures with torch_compilable_check are not the same bug." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "GLM-4V model loading and tokenizer loading for est-roberta are unrelated." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "Bark voice_preset and progressive generation with inputs_embeds/past_key_values are different features with different failures." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer crashes on NER/padding vs export failures from torch_compilable_check are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "Unexpected init_empty_weights argument and strict config rejection are different loading-time issues." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "Both are model bugs, but one is offload/device mismatch and the other is DAC.from_latents mismatch; not the same code path." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "MIMI batch-vs-single padding-mask behavior and Perceiver resolution handling are different bugs." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation with cached embeddings and Qwen3 auto device-map CUDA asserts are unrelated." + }, + { + "left": "issue:44610", + "right": "issue:45325", + "accept": false, + "reason": "Processor output resolution mismatch and Qwen2.5-VL rope-index scaling are different model-specific issues." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "lm_head weight tying and bfloat16 dtype mismatches are unrelated." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "Weight-tying bug vs torch.compile/SDPA failure; no shared concrete bug." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "MIMI padding-mask correctness and OmDet-Turbo image-size mismatch are different subsystems." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "Documentation/example errors for SigLIP2 and multi-image finetuning failures in Qwen2.5-VL are separate reports." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "FSDP CPU RAM-efficient loading is not the same as non-persistent buffer initialization junk." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both touch serialization/save_pretrained behavior, but one is lm_head tying on device and the other is incorrect visual encoder keys in Qwen3.5." + }, + { + "left": "issue:43296", + "right": "issue:44991", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and tokenizer loading for est-roberta are unrelated." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "Weight tying in a model head is unrelated to BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:43296", + "right": "issue:44877", + "accept": false, + "reason": "PaddleOCR-VL load failure and strict granite_speech config rejection are different failures." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Auto device-map CUDA assert and Llava Next segmentation fault are both loading failures but not the same bug." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Qwen3 auto device mapping crash and Qwen3.5 flash-attention generation bug are distinct issues." + }, + { + "left": "issue:43792", + "right": "issue:44683", + "accept": false, + "reason": "Whisper runtime failure and compiled flex_attention failure on torch>=2.9 are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Mistral3 lm_head tying and Qwen3.5 visual encoder key serialization are different model-specific regressions." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer/padding crashes and bfloat16 dtype mismatches do not share 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": 2, + "component_count": 13, + "review_unit_index": 10, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:30064", + "issue:33357", + "issue:34567", + "issue:34634", + "issue:35707", + "issue:36010", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43316", + "issue:43317", + "issue:43322", + "issue:43388", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43708", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:43881", + "issue:44186", + "issue:44292", + "issue:44361", + "issue:44423", + "issue:44479", + "issue:44514", + "issue:44545", + "issue:44610", + "issue:44671", + "issue:44743", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44918", + "issue:44991", + "issue:45059", + "issue:45161", + "issue:45406", + "issue:45464" + ], + "soft_pairs": [ + "issue:44423|issue:45406", + "issue:34567|issue:40990", + "issue:43526|issue:43881", + "issue:43873|issue:44756", + "issue:43577|issue:43708", + "issue:42617|issue:42915", + "issue:43550|issue:43720", + "issue:44610|issue:44898", + "issue:44877|issue:44991", + "issue:44186|issue:44610", + "issue:44479|issue:44918", + "issue:36331|issue:42915", + "issue:36331|issue:42617", + "issue:44671|issue:44743", + "issue:43316|issue:44913", + "issue:43550|issue:43606", + "issue:42915|issue:43317", + "issue:41720|issue:42617", + "issue:44292|issue:44479", + "issue:41720|issue:45161", + "issue:43531|issue:43572", + "issue:44186|issue:44361", + "issue:43388|issue:45059", + "issue:43540|issue:43782", + "issue:41720|issue:43873", + "issue:42915|issue:43322", + "issue:36010|issue:42617", + "issue:45406|issue:45464", + "issue:36331|issue:41720", + "issue:44514|issue:44545", + "issue:33357|issue:35707", + "issue:43606|issue:43720", + "issue:41720|issue:43746", + "issue:30064|issue:34634" + ], + "prepared_review_unit_hash": "954bcc5436b28a3fb5fa146cd92f443b4a084ae14e781a23b54219298681e987", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37728, + "estimated_input_tokens": 9432, + "estimated_eval_tokens": 19120 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous single-bug issues with only one clear duplicate pair: Qwen2_5_VLProcessor.apply_chat_template crashes on batched input when padding=False (44514/44545). The rest look like related-sounding but distinct bugs, so I rejected the soft edges.", + "confidence": 0.86, + "canonical_issue_reason": "issue:44545 is the best canonical anchor here because it is the clearest, most concise representation of the only exact duplicate pair (with 44514) and has the stronger reference signal.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44545 is the strongest standalone issue in this set: it is specific, reproducible, and clearly duplicated by 44514, making it the best representative of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both are serve/runtime crashes in multimodal contexts, but they involve different processors and failure modes ('str' .to vs missing _tokenizer)." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "Trainer token counting and perplexity on GPT-OSS are unrelated bugs." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and GLM-4V loading failure are different code paths and symptoms." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Both touch offloading/memory, but one is quantization offload behavior and the other is mmap/OOM on Strix Halo; not the same bug." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "BLIP2 dtype propagation and Trainer resume max_steps calculation are unrelated." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "A 3d_parallel.py failure and a FineGrainedFP8Config failure are different model/runtime problems." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "Both involve model loading/execution infrastructure, but the concrete failures differ: torch.compile+SDPA vs packed weights not unpacked during accelerate loading." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "These are both image-size/resolution issues, but on different models and with different expected-processing mismatches." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config loading for granite_speech and tokenizer loading for EMBEDDIA/est-roberta are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "Tokenizer crashes on NER/padding vs processor output size mismatch are different bugs." + }, + { + "left": "issue:44479", + "right": "issue:44918", + "accept": false, + "reason": "Both mention Qwen-family models, but one is a video-input regression and the other is an embeddings/trainer failure." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "Unexpected compute_loss kwargs and FineGrainedFP8 model failure are unrelated." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Trainer loss signature mismatch is unrelated to the 3d_parallel.py failure." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "CamemBERT masked-LM predictions and Qwen modular recurrent-state reset are different model-specific bugs." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Both are config/serialization regressions, but they concern different fields and models; not the same underlying issue." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "torch.compile+SDPA on Bamba and CPU-offload device mismatch on Bark are distinct failures." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8 loading and dequantized model offload loading are both loading issues, but not the same code-path problem." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Qwen3 auto device mapping cudaErrorAssert and 3d_parallel.py failure are different problems." + }, + { + "left": "issue:44292", + "right": "issue:44479", + "accept": false, + "reason": "Qwen NVFP4 runtime error and Qwen video-input regression are distinct." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Auto device mapping cudaErrorAssert on A800 is unrelated to GPT-OSS MoE TP failure." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE and missing pad_token_idx in StableLmConfig are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Different tokenizer bugs with different models and failure mechanisms." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "gather_for_metrics label dropping and SAM3 PCS text/box behavior are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Qwen3OmniMoe video processing ValueError and Qwen3VL weight_only load error are separate multimodal issues." + }, + { + "left": "issue:41720", + "right": "issue:43873", + "accept": false, + "reason": "Both mention device/offload behavior, but one is a cudaErrorAssert during auto mapping and the other is quantization offload behavior; not the same bug." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "GenerationMixin import failure and 3d_parallel.py failure are unrelated." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and Qwen3.5 streaming API failure are different serve bugs." + }, + { + "left": "issue:36331", + "right": "issue:41720", + "accept": false, + "reason": "Trainer compute_loss signature mismatch is unrelated to Qwen3 device-mapping crashes." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "These are the same bug reported twice: identical Qwen2_5_VLProcessor.apply_chat_template batched-input crash with padding=False." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "MacOS bus error on CLIP and progressive generation with inputs_embeds/past_key_values are different issues." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "Bark CPU-offload device mismatch and BitNet packed-weight unpacking failure are not the same concrete bug." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Qwen3 device-mapping crash and GraniteSpeech PEFT checkpoint loading are unrelated." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Void segmentation-map processing and BarkProcessor voice_preset failure are unrelated model/processor bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative. The only clearly duplicated pair in the packet is 44514/44545, and choosing 44545 as canonical/best issue is supported by the stronger inbound-reference signal. The rejected soft edges are phrased cautiously and do not overclaim equivalence based on the packet." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 11, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:34567", + "issue:34634", + "issue:35707", + "issue:39692", + "issue:40444", + "issue:41720", + "issue:42886", + "issue:43381", + "issue:43421", + "issue:43531", + "issue:43550", + "issue:43596", + "issue:43606", + "issue:43645", + "issue:43653", + "issue:43746", + "issue:43782", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44062", + "issue:44077", + "issue:44186", + "issue:44220", + "issue:44368", + "issue:44451", + "issue:44492", + "issue:44509", + "issue:44589", + "issue:44610", + "issue:44625", + "issue:44671", + "issue:44843", + "issue:44898", + "issue:44918", + "issue:44936", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45183", + "issue:45356", + "issue:45357", + "issue:45381", + "issue:45479" + ], + "soft_pairs": [ + "issue:43550|issue:43653", + "issue:44492|issue:44509", + "issue:44368|issue:45357", + "issue:44625|issue:45479", + "issue:43606|issue:43653", + "issue:43421|issue:43927", + "issue:44898|issue:45072", + "issue:43782|issue:45381", + "issue:34567|issue:35707", + "issue:34634|issue:41720", + "issue:43381|issue:43844", + "issue:42886|issue:44843", + "issue:44077|issue:44625", + "issue:43421|issue:43653", + "issue:44451|issue:44991", + "issue:44062|issue:44451", + "issue:44918|issue:44936", + "issue:44987|issue:45042", + "issue:43531|issue:44589", + "issue:43994|issue:44671", + "issue:39692|issue:44610", + "issue:43421|issue:44186", + "issue:45081|issue:45356", + "issue:39692|issue:43873", + "issue:43531|issue:43645", + "issue:45183|issue:45479", + "issue:43746|issue:45381", + "issue:43596|issue:44220", + "issue:43746|issue:43873", + "issue:40444|issue:43746" + ], + "prepared_review_unit_hash": "bc1a8b98bda57bb1f946572eb9d852ea8e5a9166ca7093e552e8a93d26760fcb", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36412, + "estimated_input_tokens": 9103, + "estimated_eval_tokens": 18462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly unrelated issues from the same repo. The offline-tokenizer pair (42886 \u2194 44843) is thematically close, but with only title-level evidence it is not strong enough to call them the same bug or a mergeable duplicate, so it should be rejected as a soft edge.", + "confidence": 0.74, + "canonical_issue_reason": "issue:44843 is still the best anchor because it names a concrete suspected root cause (`model_info()` in `_patch_mistral_regex`) and directly describes the offline-loading failure mode.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44843 is the best global issue candidate because it is the most specific, actionable, and likely to subsume the broader offline tokenizer regression space among the provided issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "Different components and failure modes: torch.compile/SDPA model execution vs tokenizer special-token registration." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs/wording-related, but they address different content and different changes." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "Unrelated Qwen3.5 config warning vs incorrect visual encoder key saving; different bug paths." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "One is config propagation in Qwen3.5, the other is a generic classification loss degeneracy across models." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to a tokenizer special-token decode bug." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Both involve token IDs/special tokens, but one is runtime post-processor updates and the other is save/load losing custom IDs; not the same bug." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Different model families and different problems: image resolution handling vs bfloat16 dtype mismatch." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen VL issues, but one is `from_pretrained(weight_only=True)` loading and the other is video vision_position_ids; not the same code path." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "Trainer state token counting is unrelated to progressive generation with inputs_embeds/past_key_values." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "Bark voice_preset and Qwen3 auto device mapping crashes are unrelated." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Gradient checkpointing in eval mode is a usage/validation issue, while the other is a ZeRO-3 gradient abnormality during training." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": false, + "reason": "Title-level similarity around HF_HUB_OFFLINE/tokenizer loading is not enough to confirm the same root cause; the pair is plausible but too weak to merge as a duplicate." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "Optional `post_init` policy for patchtsmixer is unrelated to Qwen3.5 num_labels propagation." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Special-token runtime updates and BigBird mask-token registration are different tokenizer bugs." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading regressions, but they affect different models and likely different root causes." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "`AddedToken` argument collision is not the same as model-specific tokenizer loading failure." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "TRL SFT unpacking with inputs_embeds is unrelated to trainer.evaluate() failing after train()." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "Model-loading failure and PIL image-processor torchvision dependency are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to missing Float8 storage support." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different models and different incorrect-output bugs; too broad to merge." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Both are vision/documentation-ish, but the concrete failures and model/processor mismatches are different." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Tokenizer special-token update logic is unrelated to LayoutLMv2 NER padding/truncation crashes." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Both mention `fix_mistral_regex`, but one is an AttributeError crash and the other is a codec-handling regression; not enough evidence they are the same underlying bug." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 doc/example issues are unrelated to quantization/offloading behavior." + }, + { + "left": "issue:43531", + "right": "issue:43645", + "accept": false, + "reason": "Sliding-window MoE behavior is unrelated to custom-model initialization in Jupyter notebooks." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Input validation errors are unrelated to the zero-loss classification bug." + }, + { + "left": "issue:43746", + "right": "issue:45381", + "accept": false, + "reason": "PEFT checkpoint loading and Qwen2.5-VL video position IDs are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Deepspeed/BertModel zero3 init bug is unrelated to fbank feature extraction." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "PEFT checkpoint loading and quantization offload behavior are different problems." + }, + { + "left": "issue:40444", + "right": "issue:43746", + "accept": false, + "reason": "Multi-image iterable dataset finetuning and GraniteSpeech PEFT checkpoint loading are unrelated." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The summary is mostly grounded, but the soft-edge acceptance of issue:42886 \u2194 issue:44843 is too strong for the packet. The packet only gives title-level evidence, and while both mention HF_HUB_OFFLINE/tokenizer loading, there isn\u2019t enough here to confidently call them the same bug or a mergeable duplicate. A more conservative verdict would reject that pair unless the issue bodies provide stronger shared-root-cause evidence." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 12, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:43335", + "issue:43421", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43618", + "issue:43643", + "issue:43644", + "issue:43650", + "issue:43653", + "issue:43697", + "issue:43716", + "issue:43720", + "issue:43749", + "issue:43756", + "issue:43782", + "issue:43818", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43906", + "issue:43927", + "issue:43994", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44442", + "issue:44448", + "issue:44488", + "issue:44561", + "issue:44661", + "issue:44671", + "issue:44811", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45381" + ], + "soft_pairs": [ + "issue:43540|issue:44008", + "issue:43618|issue:43697", + "issue:43720|issue:43819", + "issue:43716|issue:44008", + "issue:44220|issue:44661", + "issue:43335|issue:43756", + "issue:43653|issue:43720", + "issue:43421|issue:44811", + "issue:44186|issue:44898", + "issue:44843|issue:45081", + "issue:43550|issue:45072", + "issue:30064|issue:36010", + "issue:43577|issue:43818", + "issue:43550|issue:43927", + "issue:44062|issue:44488", + "issue:43540|issue:45381", + "issue:43643|issue:44877", + "issue:43577|issue:43582", + "issue:43531|issue:43643", + "issue:43782|issue:44977", + "issue:43644|issue:43749", + "issue:43653|issue:43819", + "issue:44448|issue:44671", + "issue:44561|issue:45020", + "issue:43825|issue:43994", + "issue:43550|issue:43819", + "issue:44186|issue:44442", + "issue:43825|issue:43827", + "issue:43650|issue:43906", + "issue:44821|issue:45084" + ], + "prepared_review_unit_hash": "b5b13790f087834893d5125774efa5130e85253d9da426f3f49d7c3571dc85d7", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36742, + "estimated_input_tokens": 9186, + "estimated_eval_tokens": 18628 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a broad mix of unrelated Transformers issues across different models, tokenizer/config, multimodal, and loading paths. The soft-similarity pairs are mostly topical overlaps, not the same underlying bug or change, so none should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Different failures and code paths: Qwen3OmniMoe video processing ValueError vs Gemma3n forward attribute collision." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "CLIPOutput attentions missing and RTDetrV2 output drift are unrelated model behaviors." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "Packed-weight loading/unpacking in BitNet is unrelated to DAC.from_latents vs forward STE mismatch." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Both are multimodal, but one is image preprocessor/model dtype mismatch and the other is a Gemma3n variable-name bug." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio feature extraction issue vs tokenizer mapping/add-new-model failure; different subsystems." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction bug and Smollm3 RoPE layer dropping are distinct config/model issues." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "Tokenizer special-token registration bug is unrelated to BitNet packed-weight loading." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Runtime post-processor update for special tokens differs from Whisper batch_decode skip_special_tokens behavior." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "LayoutLMv2 tokenizer padding/NER crash is not the same as Perceiver image classification with non-default resolution." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Both involve Mistral regex helpers, but one is offline model_info access and the other is a backend_tokenizer AttributeError; not the same bug." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Image processor segmentation-map handling is unrelated to the GenerationMixin import error." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "Blip2 dtype persistence and Video-LLaVA temporal-attention/weight-sharing are different issues." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "torch.compile SDPA failure and DiaConfig save/load token-ID loss are unrelated bugs." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken special-argument conflict is unrelated to loading a specific tokenizer model." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Different Qwen multimodal failures: video-input processing error vs wrong vision_position_ids in qwen2.5-vl." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "Missing fields with trust_remote_code and strict granite_speech config loading are distinct config-loading problems." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "Different dtype-related symptoms and platforms: Blip2 load dtype mismatch vs Apple Silicon allocator warmup TypeError." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Sparse-layer creation bug in SwitchTransformers is unrelated to trust_remote_code missing fields." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Qwen3VL weight_only loading error and Qwen3.5 flash-attention generation issue are different failure modes." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Non-persistent buffer junk in v5 and FSDP CPU RAM efficient loading are separate initialization/loading regressions." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Tokenizer special-token registration and DAC latent/forward mismatch are unrelated." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Pegasus v4/v5 output differences and CamemBERT masked-LM prediction issues affect different models and code paths." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Both mention remote_code/trust_remote_code, but one is the removed is_torch_fx_available API and the other is a broader remote_code regression report." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "pipeline() translation error messaging and SigLIP2 AutoModel/pipeline bad outputs are different issues." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Bamba SDPA/torch.compile failure is unrelated to DAC.from_latents vs forward mismatch." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "LayoutLMv2 tokenizer crash is not the same as AutoTokenizer failing to load FastSpeech2ConformerTokenizer." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "One is an incorrect error message, the other is stale docs; related topic but not the same concrete bug/change." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "Placeholder/nonsensical issue title versus an isolated reproduction of another bug; no shared underlying change." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor URL loading and template-node compilation errors 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": 2, + "component_count": 13, + "review_unit_index": 13, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:30064", + "issue:34634", + "issue:36010", + "issue:36331", + "issue:38175", + "issue:41720", + "issue:42175", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43452", + "issue:43454", + "issue:43526", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43638", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43746", + "issue:43749", + "issue:43782", + "issue:43819", + "issue:43854", + "issue:44060", + "issue:44291", + "issue:44368", + "issue:44485", + "issue:44509", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44671", + "issue:44805", + "issue:44811", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45406", + "issue:45488", + "issue:45538" + ], + "soft_pairs": [ + "issue:41720|issue:44991", + "issue:43606|issue:45072", + "issue:44805|issue:44936", + "issue:30064|issue:36331", + "issue:44610|issue:45072", + "issue:44485|issue:44509", + "issue:42175|issue:45081", + "issue:43782|issue:45325", + "issue:34634|issue:36010", + "issue:43720|issue:45072", + "issue:44060|issue:44368", + "issue:43262|issue:43819", + "issue:45081|issue:45406", + "issue:43749|issue:44291", + "issue:43582|issue:44857", + "issue:43653|issue:44671", + "issue:43526|issue:43697", + "issue:43526|issue:44554", + "issue:43653|issue:44912", + "issue:45084|issue:45538", + "issue:43454|issue:43577", + "issue:34634|issue:36331", + "issue:44568|issue:45488", + "issue:43257|issue:43329", + "issue:43746|issue:44811", + "issue:43638|issue:44936", + "issue:43452|issue:44991", + "issue:43854|issue:44977", + "issue:38175|issue:44610" + ], + "prepared_review_unit_hash": "6d93d5f0d88fbbfe741d2497c3f303007c5af952a86f8a5d4dc18d3d1a779bad", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36957, + "estimated_input_tokens": 9240, + "estimated_eval_tokens": 18736 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The soft pairs are all cross-topic regressions affecting different models, APIs, or code paths; none look like true duplicates. Most are superficially similar only at the level of broad subsystem (tokenizers, loading, dtype, processors), but the concrete failures differ.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:45081 is the strongest standalone report here: it has a clear traceback, a specific failing code path (`_patch_mistral_regex` / `backend_tokenizer`), and inbound references, making it the most actionable issue among the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Different failures: Qwen3 auto device mapping triggers `cudaErrorAssert`, while the other is a tokenizer load failure for a different model family." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "Both involve inference-time mismatches, but one is a Bark CPU-offload device mismatch and the other is a bf16 dtype mismatch in different model code paths." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "Mask shape/indexing error during model use is unrelated to `trainer.evaluate()` failing after `trainer.train()`." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "Image processor void-segmentation handling and a custom trainer signature change are unrelated bugs." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Processor image-size mismatch for OmDet-Turbo is not the same bug as dtype mismatches in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion is unrelated to documentation still mentioning removed pipeline tasks." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Package extras/backend installation issue is unrelated to a tokenizer regex patch crash." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen vision-language bugs: one is a load error with `weight_only=True`, the other is incorrect RoPE position scaling." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "A BarkProcessor `voice_preset` problem is not the same as a missing `GenerationMixin` import." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "Packed-weight unpacking during accelerate loading is a different loading bug than bf16 dtype mismatches at inference." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention tied-weights warnings, but they describe different models and different root causes." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and DAC forward/from_latents mismatch are unrelated." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "The Mistral tokenizer regex patch crash is unrelated to Gemma4Processor missing `_tokenizer` in transformers serve." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "Both concern loading flows, but one is FSDP CPU RAM efficient loading and the other is an `init_empty_weights` argument regression; not the same code-path problem." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError and CUDA float16 loss crash are different platform/model issues." + }, + { + "left": "issue:43653", + "right": "issue:44671", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and CamemBERT masked-LM prediction regressions are different failures." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "BEiT label reduction bug and RTDetrV2 output changes in v5 are unrelated vision issues." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "Image processor label reduction bug is unrelated to the MPS attention correctness issue." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "Tokenizer special-token decoding bug and MXFP4 quantization fallback are unrelated." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Template compilation error and CLIPTokenizer `model_max_length` are unrelated tokenizer/config issues." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "AyaVision tied-weight loading and Blip2 dtype persistence are different model-loading bugs." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "Repeated pair: BarkProcessor preset issue and trainer loss-signature issue are unrelated." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both are tokenizer regressions, but one is missing BOS/EOS addition and the other overrides ByteLevel pre-tokenizer behavior; different concrete defects." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "MOE weight conversion during accelerate+deepspeed loading is unrelated to the undefined video-token counting helpers." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "PEFT local checkpoint loading and Whisper `batch_decode()` special-token handling are different bugs." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Deepspeed ZeRO-3 empty tensor indexing and `trainer.evaluate()` breaking after training are unrelated training-flow issues." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "GGUF loading support and tokenizer loading for `EMBEDDIA/est-roberta` are separate `from_pretrained` failures." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "GLM-4.7-Flash unit-test loading and Qwen3.5 flash-attention generation problems are different model-specific bugs." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Both are vision-model regressions, but one is zero probabilities in SigLIP2 and the other is a processor/model input-size mismatch." + } + ] + }, + "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": 13, + "review_unit_index": 14, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:33357", + "issue:34567", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43317", + "issue:43329", + "issue:43450", + "issue:43452", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43645", + "issue:43696", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43819", + "issue:43927", + "issue:43931", + "issue:44291", + "issue:44479", + "issue:44560", + "issue:44568", + "issue:44589", + "issue:44610", + "issue:44756", + "issue:44811", + "issue:44912", + "issue:44938", + "issue:44987", + "issue:45042", + "issue:45072", + "issue:45137", + "issue:45356", + "issue:45357", + "issue:45381", + "issue:45405", + "issue:45538" + ], + "soft_pairs": [ + "issue:44589|issue:45042", + "issue:39692|issue:44912", + "issue:43257|issue:43696", + "issue:43819|issue:43927", + "issue:43262|issue:44811", + "issue:43720|issue:44610", + "issue:43550|issue:44912", + "issue:36010|issue:43645", + "issue:44568|issue:45356", + "issue:43749|issue:44756", + "issue:43329|issue:44479", + "issue:43531|issue:45042", + "issue:43257|issue:43931", + "issue:36010|issue:36331", + "issue:43257|issue:43756", + "issue:43329|issue:43450", + "issue:43577|issue:45072", + "issue:43697|issue:45357", + "issue:42175|issue:44938", + "issue:43452|issue:43582", + "issue:41720|issue:43540", + "issue:44479|issue:45381", + "issue:43927|issue:44912", + "issue:33357|issue:34567", + "issue:42915|issue:43742", + "issue:44987|issue:45538", + "issue:44291|issue:45042", + "issue:43317|issue:45538", + "issue:43540|issue:44560", + "issue:45137|issue:45405" + ], + "prepared_review_unit_hash": "487e66e3dbfa979f457d3e0c48ed4a6d0ad690e2d815355e593a3609affce0ad", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36778, + "estimated_input_tokens": 9195, + "estimated_eval_tokens": 18646 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of unrelated issues across loading, tokenizer, multimodal/video, training, dtype, and packaging bugs. There is one close Qwen video-regression subfamily, but the rest are not duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 44479 is the broadest representative of the only clearly related subgroup (Qwen video-input regressions), but the overall cluster is not a true duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44479 is the best representative issue because it captures the widest blast radius among the video-regression reports, though it still does not subsume the rest of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Different bug classes: float8 storage/load failure vs PIL backend dependency checking." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "Doc-example/model-processor mismatch and quantization failure are unrelated to MXFP4 loading fallback." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "Qwen3 MoE weight-conversion bug and GPT-oss OOM are different problems." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "DAC latent reconstruction and DiaConfig token-ID persistence are unrelated code paths." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Audio chat-template sampling-rate default and Whisper batch_decode skip_special_tokens are different processor bugs." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "Packed-weight unloading during accelerate load is unrelated to an OmDet-Turbo image-size mismatch." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "torch.compile/SDPA failure and MXFP4 quantization loading are separate issues." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import breakage is not the same as Jupyter custom-model initialization." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Tokenizer special-token insertion and codec-handling/warning regression are distinct tokenizer bugs." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and Strix Halo mmap OOM are different memory-loading problems." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Undefined video-helper references are a different failure mode from the broader Qwen video-input regression." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior and PIL/torchvision dependency handling are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "MoE weight conversion and Qwen3-VL weight-shape mismatch are not the same bug." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "Importing GenerationMixin and CustomTrainer.compute_loss signature changes are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Qwen3 MoE accelerate loading and Smollm3 RoPE-layer dropping are different model issues." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Undefined multimodal-token helpers and batched video-shape errors are separate video-processing bugs." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Blip2 dtype sticking in float32 and SwitchTransformers/TimmWrapper dtype mismatches are distinct." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "RTDetrV2 inference divergence and Qwen3.5 save_pretrained key mapping are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Packaging missing TensorFlow and Python 3.14 import/load failure are different issues." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "gguf_file loading breakage and Apple Silicon warmup TypeError are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Auto device-mapping CUDA assert and Qwen3OmniMoe video-processing ValueError are different code paths." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "These describe the same Qwen video-input regression family: both center on incorrect video position-id handling in 5.3.0." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig custom-token persistence and MXFP4 loading fallback are unrelated." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "MacOS bus error with CLIP and TrainerState token counting are completely unrelated." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and MobileLLM key error are different loading bugs." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Model loading regression for physical-intelligence/fast and CLIPTokenizer max_length bug are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "init_empty_weights argument compatibility and PIL backend torchvision requirement are different bugs." + }, + { + "left": "issue:43317", + "right": "issue:45538", + "accept": false, + "reason": "Device-map/offload dequantization failure and CLIPTokenizer model_max_length are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Both are video-related, but Qwen3OmniMoe ValueError and Qwen3-vl-embedding StopIteration are different model/code-path failures." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque error and PEFT minimum-version bump are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative: it treats the cluster as mostly unrelated while recognizing the one plausible Qwen video-regression subgroup. The accepted soft edge (44479/45381) is supported by the titles, and the rejected pairs do not overclaim duplication based on the available evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 15, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:28282", + "issue:29127", + "issue:34567", + "issue:34689", + "issue:36010", + "issue:36331", + "issue:38175", + "issue:42175", + "issue:42915", + "issue:43452", + "issue:43493", + "issue:43526", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43643", + "issue:43646", + "issue:43688", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43782", + "issue:43824", + "issue:43906", + "issue:43994", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44484", + "issue:44488", + "issue:44514", + "issue:44561", + "issue:44589", + "issue:44704", + "issue:44792", + "issue:44821", + "issue:44898", + "issue:44964", + "issue:44991", + "issue:45081", + "issue:45216", + "issue:45290", + "issue:45325", + "issue:45375", + "issue:45381", + "issue:45479", + "issue:45538" + ], + "soft_pairs": [ + "issue:43906|issue:45381", + "issue:43540|issue:44479", + "issue:42915|issue:45538", + "issue:42915|issue:44821", + "issue:42175|issue:43824", + "issue:43526|issue:45479", + "issue:42175|issue:44488", + "issue:44062|issue:44991", + "issue:38175|issue:44898", + "issue:28282|issue:34689", + "issue:43646|issue:43749", + "issue:43582|issue:43782", + "issue:28282|issue:29127", + "issue:43493|issue:43994", + "issue:43742|issue:44291", + "issue:43643|issue:45375", + "issue:43756|issue:45325", + "issue:36010|issue:42175", + "issue:44484|issue:44704", + "issue:34567|issue:36331", + "issue:44291|issue:44821", + "issue:44792|issue:44964", + "issue:43452|issue:44488", + "issue:44062|issue:45081", + "issue:43688|issue:43749", + "issue:44514|issue:45290", + "issue:43643|issue:44561", + "issue:43452|issue:43577", + "issue:43742|issue:44589", + "issue:44514|issue:45381", + "issue:43697|issue:45216" + ], + "prepared_review_unit_hash": "59ad27e1c4710c10d6515dc43c55c21d6112e7f0adce329efbe367e6a05ea629", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37231, + "estimated_input_tokens": 9308, + "estimated_eval_tokens": 18872 + }, + "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 and model-specific loading bugs. Only a couple of pairs look like they share the same underlying code path; most should stay separate.", + "confidence": 0.39, + "canonical_issue_reason": "Issue 43646 is the broadest umbrella item here: a v5.0 custom-model initialization regression. It best matches the cluster\u2019s recurring theme of version-related loading/init breakages, even though many other issues are too specific to be true duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "43646 is the most representative issue for this cluster because it is the most general and central failure mode among the mostly version-regression/loading reports. None of the other issues is as good an umbrella candidate.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43906", + "right": "issue:45381", + "accept": false, + "reason": "Both mention Qwen2.5-VL-style regressions, but one is an isolated reproduction of a different issue and the other is a specific v5.3.0 position-id bug. Not the same underlying problem." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Both are video-input problems in Qwen-family models, but they involve different model families and different symptoms. Too broad to merge." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "Qwen3Moe FP8 failure and CLIPTokenizer max_length are unrelated bugs." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Different subsystems and different failures; no shared underlying bug." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "One is backend/package installation, the other is a model import failure. Not the same issue." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "BEiT label reduction and zero-loss classification are different code paths and model families." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Packaging/backend availability versus loading a specific model; unrelated causes." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Different tokenizer/loading failures with no clear shared root cause." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "SigLIP2 zero-probabilities and Perceiver interpolation failure are different models and bugs." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "Missing PyTorch dependency and Llama 3.2 model loading regression are not the same problem." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Generic custom-model init breakage and FSDP CPU RAM-efficient loading are different loading paths." + }, + { + "left": "issue:43582", + "right": "issue:43782", + "accept": false, + "reason": "Apple Silicon allocator TypeError and Qwen3VL weight_only loading error are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Environment dependency error versus a LayoutLMv3 error-message-quality issue." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Both target google/siglip2-base-patch16-224 and describe incorrect outputs from the HF SigLIP2 implementation; the JAX discrepancy and nonsensical pipeline results look like the same underlying model bug." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Different load-time TypeErrors with different triggers and code paths." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both involve config-field handling, but one is a generic trust_remote_code config-loss issue and the other is a Qwen3_5MoeVisionConfig field omission. Too different." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen-related, but one is about SmolLM3 RoPE layers and the other is a Qwen2.5-VL position-id regression." + }, + { + "left": "issue:36010", + "right": "issue:42175", + "accept": false, + "reason": "GenerationMixin import failure and TensorFlow backend packaging are unrelated." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "save_pretrained shard-size semantics and AutoProcessor cached_file kwargs are different API bugs." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "TrainerState token counting and custom Trainer compute_loss signature are separate trainer issues." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "init_empty_weights argument handling and AutoImageProcessor URL loading are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Janus test failure and Phi-4 multimodal load failure are both multimodal, but not the same concrete bug." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "gguf_file handling and loading cjvt/sleng-bert are different model-loading failures." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "AddedToken keyword conflict and Mistral regex patch crash are different tokenizer bugs." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Auxiliary-loss normalization and FSDP CPU RAM-efficient loading are unrelated." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both are apply_chat_template crashes, but the triggers differ: batched padding=False versus assistant tool-call messages with no content." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both involve trust_remote_code/custom-model behavior, but one is missing config fields and the other is removal of is_torch_fx_available. Different root causes." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "Different loading bugs: gguf_file handling versus BLIP2 dtype propagation." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "A key error when loading MobileLLM and a missing Float8 storage type are different failures." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Both mention Qwen2.5-VL chat/image/video behavior, but one is chat-template batching and the other is a vision-position-id regression. Not the same bug." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 save_pretrained corruption are separate regressions." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The single accepted pair (issue:43493 / issue:43994) is at least plausibly the same SigLIP2 bug, and the rejections avoid overclaiming duplicate status for unrelated model- or subsystem-specific failures. The choice of issue:43646 as the umbrella issue is also defensible given the cluster\u2019s strong loading/init regression theme." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 16, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:30064", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:35141", + "issue:38175", + "issue:40444", + "issue:41720", + "issue:41762", + "issue:43122", + "issue:43295", + "issue:43452", + "issue:43525", + "issue:43531", + "issue:43577", + "issue:43643", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43819", + "issue:43906", + "issue:43927", + "issue:44361", + "issue:44442", + "issue:44464", + "issue:44466", + "issue:44479", + "issue:44568", + "issue:44625", + "issue:44743", + "issue:44811", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45005", + "issue:45042", + "issue:45081", + "issue:45356", + "issue:45440" + ], + "soft_pairs": [ + "issue:45042|issue:45081", + "issue:43531|issue:44821", + "issue:44625|issue:44877", + "issue:43819|issue:44811", + "issue:43720|issue:44898", + "issue:43688|issue:45440", + "issue:43653|issue:44568", + "issue:43452|issue:44843", + "issue:44466|issue:45005", + "issue:44464|issue:44743", + "issue:43295|issue:45356", + "issue:41762|issue:44936", + "issue:24643|issue:29127", + "issue:41720|issue:43906", + "issue:40444|issue:44811", + "issue:43525|issue:43643", + "issue:44361|issue:44442", + "issue:38175|issue:43295", + "issue:43577|issue:43697", + "issue:43653|issue:44361", + "issue:44843|issue:44991", + "issue:28282|issue:32090", + "issue:28282|issue:35141", + "issue:28282|issue:33357", + "issue:44479|issue:44936", + "issue:43122|issue:43906", + "issue:44977|issue:45356", + "issue:30064|issue:33666", + "issue:44464|issue:44977", + "issue:43653|issue:43927" + ], + "prepared_review_unit_hash": "8b660bb64fd7ed59c709b4f732b0fc477f9b51faf553aa3468fabdfddd4e6064", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36469, + "estimated_input_tokens": 9118, + "estimated_eval_tokens": 18492 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated issues across different models and subsystems. None of the soft edges are strong enough to merge as true duplicates; at most they share broad themes like tokenization, config loading, or generation regressions.", + "confidence": 0.62, + "canonical_issue_reason": "No single canonical issue clearly covers the set; if forced, issue:44843 is the closest hub because it describes a concrete tokenizer-loading regression in a narrow code path that is easy to triage.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44843 is the strongest standalone issue: it has a specific failure mode, a narrow affected path (_patch_mistral_regex), and a clear regression trigger (HF_HUB_OFFLINE).", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Different subsystems: PIL/image-processor torchvision dependency vs a Mistral tokenizer regex crash." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Both involve loading behavior, but one is Qwen3-MoE sliding-window logic and the other is AutoImageProcessor URL loading." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Both are config-related, but one is num_labels propagation and the other is strict config loading for granite_speech." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "Unrelated code paths: DAC latent conversion vs Whisper processor batch_decode behavior." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "Different models and bugs: packed-weight loading for BitNet vs Perceiver image-classification resolution handling." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Both are MoE-related, but they concern different models and different failure modes, not the same bug." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Both touch special tokens, but one is BigBird mask-token registration and the other is BOS/EOS insertion for a different tokenizer." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "Both affect tokenizer/model loading, but one is gguf_file handling and the other is an offline Mistral regex patch regression." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": false, + "reason": "Both mention tied weights, but one is device-dependent serialization and the other is translation-model behavior in v5." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Both are generation regressions, but compiled-forward chunking and recurrent-state reset are different concrete bugs." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Both are regressions, but one is processor/tokenizer access in custom code and the other is Kimi-K2.5 codec handling." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Different failures: DeepSpeed ZeRO-3 load-time IndexError vs trainer.evaluate() failing after training." + }, + { + "left": "issue:24643", + "right": "issue:29127", + "accept": false, + "reason": "DeepSpeed training RuntimeError and LayoutLMv3 box-validation messaging are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:43906", + "accept": false, + "reason": "Auto device mapping cuda assert and an isolated reproduction of another issue do not match the same underlying bug." + }, + { + "left": "issue:40444", + "right": "issue:44811", + "accept": false, + "reason": "Different areas: multi-image IterableDataset finetuning vs Whisper batch_decode special-token handling." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "Related only at a high level; one is a specific Llama4Config pad_token_id failure, the other is generic trust_remote_code config hydration." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "Different tokenizer failures with different affected components and error surfaces." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 is not the same bug as the processor/tokenizer regression in v4.57.5." + }, + { + "left": "issue:43577", + "right": "issue:43697", + "accept": false, + "reason": "BLIP-2 dtype propagation and RTDetrV2 output drift are separate model correctness issues." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "Tokenizer special-token registration and MLukeTokenizer AttributeError are unrelated." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-load regressions, but one is offline Mistral regex patching and the other is est-roberta loading on v5." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "Missing-PyTorch import error and Trainer GPU broadcast TypeError are different problems." + }, + { + "left": "issue:28282", + "right": "issue:35141", + "accept": false, + "reason": "Import-time missing dependency vs tied-weight reinitialization are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "Missing-PyTorch ImportError does not match the MacOS bus-error CLIP regression." + }, + { + "left": "issue:44479", + "right": "issue:44936", + "accept": false, + "reason": "Qwen video-input regression and trainer.evaluate() failure are different issues." + }, + { + "left": "issue:43122", + "right": "issue:43906", + "accept": false, + "reason": "Both mention tokenizer behavior, but one is a version-to-version tokenization change and the other is only an isolated reproduction of a different issue." + }, + { + "left": "issue:44977", + "right": "issue:45356", + "accept": false, + "reason": "Flash-attention generation failure and tokenizer codec regression are separate problems." + }, + { + "left": "issue:30064", + "right": "issue:33666", + "accept": false, + "reason": "Void segmentation-map processing and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency is not the same as the flash-attention generation bug." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "BigBird special-token registration and DiaConfig token-ID loss are different tokenizer/config bugs." + } + ] + }, + "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": 13, + "review_unit_index": 17, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:32090", + "issue:33666", + "issue:34567", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:43012", + "issue:43257", + "issue:43262", + "issue:43278", + "issue:43299", + "issue:43329", + "issue:43334", + "issue:43452", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43688", + "issue:43716", + "issue:43720", + "issue:43742", + "issue:43756", + "issue:43825", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:44361", + "issue:44451", + "issue:44464", + "issue:44479", + "issue:44554", + "issue:44610", + "issue:44625", + "issue:44821", + "issue:44849", + "issue:44913", + "issue:45084", + "issue:45200", + "issue:45229", + "issue:45325", + "issue:45357" + ], + "soft_pairs": [ + "issue:43329|issue:45325", + "issue:43531|issue:43742", + "issue:43577|issue:44554", + "issue:44479|issue:44625", + "issue:43550|issue:44361", + "issue:41720|issue:44464", + "issue:43012|issue:43577", + "issue:43504|issue:43720", + "issue:43504|issue:43550", + "issue:32090|issue:33666", + "issue:32090|issue:34567", + "issue:43262|issue:45084", + "issue:43504|issue:44610", + "issue:43278|issue:43688", + "issue:43452|issue:44451", + "issue:45200|issue:45229", + "issue:44849|issue:45357", + "issue:43334|issue:43572", + "issue:43756|issue:44913", + "issue:43493|issue:43825", + "issue:43540|issue:43716", + "issue:33666|issue:36010", + "issue:43742|issue:44821", + "issue:43525|issue:43572", + "issue:43720|issue:43927", + "issue:43606|issue:43927", + "issue:43688|issue:43873", + "issue:43278|issue:43844", + "issue:43257|issue:43299", + "issue:43720|issue:44361", + "issue:39692|issue:43493" + ], + "prepared_review_unit_hash": "7d4c231fc819e85be1cfcb4fbce56b25a80d59f99444578850d22187cb606d99", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36444, + "estimated_input_tokens": 9111, + "estimated_eval_tokens": 18478 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the soft links mostly connect superficially similar bug reports (config/token-ID regressions, multimodal video issues, dtype/offload problems) rather than the same underlying defect. I reject all soft edges. If one issue must be used as a representative, `issue:43525` is the cleanest self-contained regression report in the set.", + "confidence": 0.83, + "canonical_issue_reason": "No true canonical duplicate stands out; if forced to pick one representative, `issue:43525` is the most concrete and self-contained config regression (`pad_token_id` missing on `Llama4Config`).", + "canonical_pr_reason": null, + "best_issue_reason": "`issue:43525` is the strongest representative of the cluster\u2019s small config-regression subset: clear symptom, narrow scope, and easy to understand impact.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43329", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen VL/video-related, but one is an undefined-variable bug in multimodal token counting and the other is a rope/position-id scaling regression; different code paths." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE and a MobileLLM key error are unrelated model-loading/runtime bugs." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "One is a training/eval dtype mismatch, the other is an MPS attention shape/correctness issue; not the same defect." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "A video-input regression across Qwen VL variants is distinct from `num_labels` propagation in Qwen3.5." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Bamba `torch.compile`/SDPA failure and MLuke tokenizer attribute errors are different subsystems and symptoms." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "CUDA assert with auto device mapping is not the same as inconsistent outputs from compiled chunked generation." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "Both mention dtype, but one is a PyTorch warning during compile and the other is a Blip2 loading dtype bug." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Legacy-field loading failure in BEiT and unpacking packed weights in BitNet are different loading bugs." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Unrelated model-loading bug vs compile/SDPA runtime failure." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Trainer broadcast error and Qwen2-VL multi-GPU training are too broad to be the same concrete bug." + }, + { + "left": "issue:32090", + "right": "issue:34567", + "accept": false, + "reason": "Different Trainer issues: GPU broadcast type error vs token-count bookkeeping not updating." + }, + { + "left": "issue:43262", + "right": "issue:45084", + "accept": false, + "reason": "Audio chat-template sampling-rate default and template-node compilation errors are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "BEiT legacy config loading and OmDet-Turbo processor/model size mismatch are distinct bugs." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Training/eval dtype drift is unrelated to auxiliary-loss normalization." + }, + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "A `gguf_file` path handling bug and a specific model load failure for ScandiBERT are not the same issue." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Gemma4 text-only token-type defaults and multi-GPU CUDA OOM are different problems." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "`output_hidden_states` behavior and incorrect visual-encoder keys in `save_pretrained` are separate regressions." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Both are missing token-id fields in configs, but they affect different models/attributes and do not share a concrete code path." + }, + { + "left": "issue:43756", + "right": "issue:44913", + "accept": false, + "reason": "RoPE-layer count mismatch and `rotary_pct` reset on reload are different configuration bugs." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and a pipeline translation error message are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:43716", + "accept": false, + "reason": "Qwen3OmniMoe video processing failure and Mistral-3 image-preprocessor dtype mismatch are different multimodal bugs." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Qwen2-VL multi-GPU training and a `GenerationMixin` import error have no concrete overlap." + }, + { + "left": "issue:43742", + "right": "issue:44821", + "accept": false, + "reason": "MobileLLM key error and loading `AutoImageProcessor` from URL are unrelated loading failures." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Both involve missing token-id fields, but the missing attribute and model/config context differ; not the same underlying bug." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "BitNet packed-weight unpacking on accelerate load and DiaConfig token-ID persistence are separate failures." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to save/load loss of custom token IDs." + }, + { + "left": "issue:43688", + "right": "issue:43873", + "accept": false, + "reason": "Aux-loss normalization and quantized offloading are different behaviors in different code paths." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Embedding dtype drift and ZeRO-3 gradient amplification are not the same bug." + }, + { + "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 transformers-dev regression for Qwen3VL MoE; too different to merge." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "BitNet weight unpacking and MLuke tokenizer attribute errors are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "SigLIP2 docs-example failures and SigLIP2 HF-vs-JAX numerical discrepancy are different issues, 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": 2, + "component_count": 13, + "review_unit_index": 18, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:15354", + "issue:33357", + "issue:33666", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:43257", + "issue:43278", + "issue:43295", + "issue:43329", + "issue:43493", + "issue:43504", + "issue:43540", + "issue:43572", + "issue:43575", + "issue:43582", + "issue:43606", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43701", + "issue:43746", + "issue:43756", + "issue:43824", + "issue:43844", + "issue:43927", + "issue:43994", + "issue:44060", + "issue:44112", + "issue:44291", + "issue:44361", + "issue:44423", + "issue:44466", + "issue:44488", + "issue:44493", + "issue:44521", + "issue:44554", + "issue:44589", + "issue:44734", + "issue:44792", + "issue:44843", + "issue:44964", + "issue:44977", + "issue:45084", + "issue:45290", + "issue:45405", + "issue:45446", + "issue:45520" + ], + "soft_pairs": [ + "issue:44964|issue:45405", + "issue:43540|issue:44977", + "issue:43329|issue:43606", + "issue:33666|issue:36331", + "issue:44488|issue:44843", + "issue:43493|issue:43844", + "issue:44589|issue:45520", + "issue:43756|issue:44060", + "issue:43646|issue:44466", + "issue:43688|issue:43746", + "issue:44521|issue:45290", + "issue:44423|issue:44734", + "issue:33357|issue:33666", + "issue:43824|issue:44589", + "issue:39692|issue:43295", + "issue:43994|issue:44493", + "issue:43504|issue:43606", + "issue:43927|issue:44361", + "issue:43504|issue:43653", + "issue:43575|issue:45405", + "issue:43572|issue:44291", + "issue:43278|issue:43701", + "issue:44792|issue:45405", + "issue:43257|issue:45405", + "issue:43606|issue:44112", + "issue:15354|issue:38175", + "issue:44554|issue:45446", + "issue:43582|issue:44977", + "issue:45084|issue:45520" + ], + "prepared_review_unit_hash": "0fdfd3cb3c264f3a2a6c12122b9585ebab41b938b72bf1c288d2b14c372449c5", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36681, + "estimated_input_tokens": 9171, + "estimated_eval_tokens": 18598 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose semantic neighborhood of Transformer bug reports, mostly around multimodal/model-loading regressions, serving, tokenizer/chat-template behavior, and a few unrelated training/runtime issues. I do not see any soft pair that is clearly the same underlying bug/change, so I would not merge any of them as duplicates.", + "confidence": 0.88, + "canonical_issue_reason": "issue:44521 is the broadest and most central active report in this set: multimodal chat-template handling with downstream references, which makes it the best representative anchor even though the cluster overall is heterogeneous.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the best single issue to keep as a representative because it targets a high-signal multimodal API path and has the most visible downstream interest in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "One is a Phi-4 multimodal model load failure; the other is a PEFT version bump/regression. Different problems." + }, + { + "left": "issue:43540", + "right": "issue:44977", + "accept": false, + "reason": "Different models and failure modes: Qwen3OmniMoe video processing vs Qwen3.5 flash-attention generation." + }, + { + "left": "issue:43329", + "right": "issue:43606", + "accept": false, + "reason": "Undefined multimodal-token helper bug vs CPU offload device mismatch. Not the same code path." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Multi-GPU training for Qwen2-VL vs a CustomTrainer signature regression. Unrelated." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "Both are loading-related, but one is a specific model failure and the other is an offline-mode hub call regression. Too broad to merge." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 implementation discrepancy vs a DeepSpeed/ZeRO-3 gradient issue. Different subsystems and symptoms." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Float8 storage lookup failure vs flash_attn import error on Python 3.13. Unrelated runtime issues." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch vs Qwen3-Next tied-weights warning. Different model internals." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model initialization regression vs tied-weight serialization inconsistency. Both v5-related, but not the same bug." + }, + { + "left": "issue:43688", + "right": "issue:43746", + "accept": false, + "reason": "Aux-loss normalization in MoE models vs GraniteSpeech PEFT checkpoint loading. Different affected paths." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both involve apply_chat_template, but one is multimodal assistant masks and the other is tool-call content handling. Not the same defect." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Serve crashes in two different routes: multimodal continuous batching vs KV-cache continuation indexing." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "MacOS bus error on a CLIP model vs Qwen2-VL multi-GPU training. No shared underlying bug." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "ImportError for a missing model class vs a float8 storage TypeError. Different failure layers." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Docs example errors/quantization mismatch vs a regression in processor.tokenizer access and image passing. Related area, but not the same issue." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 bad outputs vs position-id key warnings across many models. Different observable bugs." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "Legacy field blocks BEiT semantic segmentation loading vs bark-small CPU offload mismatch. Unrelated." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "Config token-id persistence bug vs tokenizer AttributeError in tasks. Different objects and root causes." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "BEiT preset loading issue vs BigBirdTokenizer special-token registration bug. Not the same defect." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "TP OOM when loading a large Qwen model vs a PEFT minimum-version bump. Completely different." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Missing pad_token_idx in one config vs an init_empty_weights signature regression. Both 5.0-era loading issues, but not the same bug." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Embedding dtype drift in evaluate vs resume_from_checkpoint key mismatch. Different training lifecycle problems." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure vs PEFT version requirement. No overlap." + }, + { + "left": "issue:43257", + "right": "issue:45405", + "accept": false, + "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed vs a dependency version bump. Unrelated." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "Both mention device-related CI behavior, but one is a real CPU-offload runtime mismatch and the other is a stale GraniteSpeech test. Not the same issue." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript GeneratorExp export failure vs unexpected zero probabilities in SigLIP2. Different eras and code paths." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "MPS attention correctness issue vs a PyTorch version gate bug in flex_attention import. Different layers." + }, + { + "left": "issue:43582", + "right": "issue:44977", + "accept": false, + "reason": "Apple Silicon allocator TypeError vs Qwen3.5 flash-attention generation bug. Unrelated." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Template compilation TypeError vs flash_attn import KeyError. Different failure mechanisms." + } + ] + }, + "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": 13, + "review_unit_index": 19, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:37428", + "issue:38175", + "issue:39401", + "issue:40990", + "issue:43232", + "issue:43278", + "issue:43295", + "issue:43334", + "issue:43388", + "issue:43450", + "issue:43493", + "issue:43525", + "issue:43526", + "issue:43540", + "issue:43572", + "issue:43575", + "issue:43701", + "issue:43824", + "issue:43844", + "issue:43856", + "issue:43873", + "issue:43957", + "issue:43975", + "issue:44315", + "issue:44442", + "issue:44464", + "issue:44466", + "issue:44514", + "issue:44521", + "issue:44554", + "issue:44568", + "issue:44589", + "issue:44610", + "issue:44625", + "issue:44792", + "issue:44849", + "issue:44898", + "issue:44936", + "issue:44964", + "issue:44987", + "issue:45072", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45325", + "issue:45356", + "issue:45412", + "issue:45563" + ], + "soft_pairs": [ + "issue:43540|issue:45200", + "issue:45072|issue:45198", + "issue:43844|issue:44964", + "issue:43232|issue:44849", + "issue:39401|issue:44568", + "issue:37428|issue:43957", + "issue:44315|issue:44849", + "issue:44554|issue:45356", + "issue:43295|issue:44610", + "issue:43334|issue:43525", + "issue:43493|issue:43701", + "issue:45200|issue:45325", + "issue:44898|issue:45412", + "issue:44466|issue:45127", + "issue:43873|issue:44464", + "issue:40990|issue:44625", + "issue:44792|issue:45563", + "issue:43388|issue:43526", + "issue:44514|issue:44521", + "issue:43844|issue:44936", + "issue:43575|issue:44464", + "issue:37428|issue:43824", + "issue:43295|issue:43450", + "issue:43572|issue:44987", + "issue:44442|issue:45198", + "issue:43295|issue:43873", + "issue:43572|issue:44589", + "issue:43824|issue:43975", + "issue:43278|issue:43856", + "issue:38175|issue:45412" + ], + "prepared_review_unit_hash": "523f235e5bf0b505ea1aac431f30765550e18e75cbacc32394f7428d9b2248d7", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36866, + "estimated_input_tokens": 9217, + "estimated_eval_tokens": 18690 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No strong duplicate cluster emerges from these items; they span unrelated bugs across tokenizers, multimodal processors, training, generation, and model loading. The most broadly representative issue is the open multimodal assistant-mask bug.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is specific, reproducible, and still open, with clear multimodal apply_chat_template failure semantics and active discussion/inbound references.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the best representative issue because it is concrete, currently open, and likely to benefit from an upstream fix without being overly model-specific.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Both involve multimodal inputs, but the concrete failures differ: video input processing ValueError vs missing mm_token_type_ids defaulting. Different code paths and fixes." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "Unrelated bugs in different subsystems/models: dtype mismatch in inference vs save_pretrained/tokenization failure." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "Different problems: ZeRO-3 gradient inflation during training vs model loading failure for Phi-4 multimodal." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "Generation kwargs update after sync_gpus is a different bug from Qwen3.5 output_hidden_states handling." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer bugs, but they affect different models and behaviors: offset_mapping vs special-token insertion." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "ImportError for a missing flash-attention helper is not the same as meta-device loading breakage." + }, + { + "left": "issue:44315", + "right": "issue:44849", + "accept": false, + "reason": "Liger Kernel application in model_init is unrelated to output_hidden_states behavior." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "Different domains entirely: MPS attention correctness vs tokenizer regression in Kimi-K2.5." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Both involve processors, but one is a regression around processor.tokenizer/image passing, the other is an image-size mismatch." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Similar symptom shape, but different model configs and likely separate missing-field fixes; not enough evidence for one underlying bug." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "SigLIP2 implementation discrepancy is unrelated to resume_from_checkpoint key mismatch." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen multimodal issues: mm_token_type_ids defaulting vs still-image temporal rope scaling." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver resolution handling is unrelated to RT-DETR memory release on deletion." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Serialization of tied lm_head weights is unrelated to LoRA merge collapse after vocab extension." + }, + { + "left": "issue:43873", + "right": "issue:44464", + "accept": false, + "reason": "Quantization/offloading behavior is unrelated to chunked generation inconsistencies under compiled forward." + }, + { + "left": "issue:40990", + "right": "issue:44625", + "accept": false, + "reason": "Perplexity regression and num_labels propagation are different bugs with different failure modes." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "Failed janus image-generation test is not the same as a stale warning in paged generate()." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Both touch labels, but one is trainer metrics batching and the other is BeitImageProcessorFast reducing labels incorrectly." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both are in apply_chat_template, but one crashes on batched padding=False while the other produces zero assistant masks; different concrete failures." + }, + { + "left": "issue:43844", + "right": "issue:44936", + "accept": false, + "reason": "Training instability with DeepSpeed is unrelated to evaluate() failing after train()." + }, + { + "left": "issue:43575", + "right": "issue:44464", + "accept": false, + "reason": "Tensor-parallel OOM during model load is unrelated to compiled-forward chunked generation output differences." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Both are ImportErrors, but for different missing symbols and different affected APIs/models." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor.tokenizer regression is not the same as batched video processor shape handling." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig is different from a broad loading failure for physical-intelligence/fast." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Different tokenizer/model failures: FastSpeech2ConformerTokenizer loading vs Wav2Vec2 save/tokenization issues." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Custom processor API regression is unrelated to offloading not working with quantization." + }, + { + "left": "issue:43572", + "right": "issue:44589", + "accept": false, + "reason": "Missing config attribute vs Float8 storage lookup failure are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Importing a missing class and incorrect detokenization are separate issues." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "BF16-to-FP32 evaluation dtype shift is unrelated to Qwen3 MoE memory usage during training." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 and RT-DETR memory release 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": 2, + "component_count": 13, + "review_unit_index": 20, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:29942", + "issue:32090", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:42915", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43335", + "issue:43425", + "issue:43450", + "issue:43493", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43819", + "issue:43824", + "issue:43825", + "issue:43827", + "issue:43901", + "issue:43931", + "issue:43994", + "issue:44112", + "issue:44351", + "issue:44360", + "issue:44448", + "issue:44462", + "issue:44493", + "issue:44561", + "issue:44568", + "issue:44610", + "issue:44743", + "issue:44779", + "issue:44849", + "issue:44857", + "issue:44991", + "issue:45003", + "issue:45083", + "issue:45127", + "issue:45245", + "issue:45290", + "issue:45310", + "issue:45381", + "issue:45478" + ], + "soft_pairs": [ + "issue:43295|issue:43994", + "issue:44360|issue:44561", + "issue:38175|issue:43493", + "issue:43827|issue:43901", + "issue:44849|issue:45003", + "issue:43653|issue:44112", + "issue:42915|issue:43824", + "issue:43450|issue:43493", + "issue:44568|issue:44991", + "issue:45127|issue:45245", + "issue:45290|issue:45381", + "issue:43450|issue:43575", + "issue:43819|issue:45083", + "issue:44462|issue:44561", + "issue:44561|issue:44779", + "issue:44610|issue:44857", + "issue:43262|issue:43450", + "issue:43819|issue:44112", + "issue:44448|issue:44743", + "issue:43606|issue:45083", + "issue:43278|issue:43825", + "issue:29942|issue:32090", + "issue:45310|issue:45478", + "issue:37428|issue:43425", + "issue:43611|issue:43931", + "issue:43335|issue:43550", + "issue:43819|issue:44493", + "issue:39692|issue:43526", + "issue:43450|issue:43526", + "issue:37428|issue:44351" + ], + "prepared_review_unit_hash": "83e2a2b27d99a538056955caeceb6debbf55197175c0c5ac0608d60878adf4b9", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36670, + "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 loose mix of unrelated Transformers issues across models, tokenizers, processors, docs, and runtime regressions. The only clear duplicate pair is the two identical Qwen3.5 Moe from_pretrained reports; everything else looks like separate bugs or feature regressions.", + "confidence": 0.96, + "canonical_issue_reason": "issue:45310 is the best canonical issue because it is the earliest clear report of the only exact duplicate pair in the cluster, with the same Qwen3.5 Moe from_pretrained failure later repeated in issue:45478.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45310 is the strongest representative issue for this cluster: it is concrete, specific, and has a true duplicate counterpart, whereas most other items are unrelated single-bug reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:43994", + "accept": false, + "reason": "Different failures: processor.tokenizer/image handling regression vs SigLIP2 nonsensical outputs." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "Unrelated topics: DSA indexer ReLU discussion vs removal of is_torch_fx_available breaking trust_remote_code." + }, + { + "left": "issue:38175", + "right": "issue:43493", + "accept": false, + "reason": "Same model family, but different concrete symptoms; not enough to treat as one bug." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs-related, but they cover different APIs and different problems." + }, + { + "left": "issue:44849", + "right": "issue:45003", + "accept": false, + "reason": "Qwen3.5 hidden-states bug is unrelated to sys.modules access in modeling_utils." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "Different components and failures: tokenizer special-token bug vs GraniteSpeech CI device override test." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "FP8 config failure in Qwen3Moe is unrelated to Qwen2.5-VL import error." + }, + { + "left": "issue:43450", + "right": "issue:43493", + "accept": false, + "reason": "Batched video shape bug is not the same as SigLIP2 implementation discrepancy." + }, + { + "left": "issue:44568", + "right": "issue:44991", + "accept": false, + "reason": "Both concern tokenizers in v5, but they are different models with different failure modes." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/tied embeddings collapse and category-count RuntimeError are unrelated bugs." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat template tool-call crash and Qwen2.5-VL video position ids are different code paths." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Video processor shape bug and tp OOM on Qwen2-57B are unrelated." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC from_latents mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." + }, + { + "left": "issue:44462", + "right": "issue:44561", + "accept": false, + "reason": "AutoTokenizer repository-loading issue is not the same as the v5 trust_remote_code regression." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5-related, but one is API removal and the other is Deepseek tokenizer output regression." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "OmDet-Turbo input-size mismatch is unrelated to LwDetrImageLoss AMP/CUDA crashing." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sample-rate default bug and video batch-shape bug are different processor issues." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC latent/STE mismatch is unrelated to GraniteSpeech CI device override flakiness." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Pegasus v4/v5 output differences are unrelated to qwen3_5 recurrent-state reset behavior." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "CPU offload device mismatch in bark-small is unrelated to qwen3_omni_moe helper behavior." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Embedding dtype drift in eval and pipeline translation error-message regression are different bugs." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 test failures are unrelated to Trainer _gpu_broadcast_one NoneType errors." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Identical title and same Qwen3.5 Moe from_pretrained failure; these are the same underlying bug." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "ImportError for flash-attention helper is not the same as torch 2.10 incompatibility." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "base_model_prefix loading breakage and Qwen3-VL shape mismatch are unrelated model-loading bugs." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug and Bamba torch.compile/SDPA failure are different issues." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "DAC latent mismatch is unrelated to position-id key regressions across models." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "SigLIP2 docs example errors are unrelated to BeitImageProcessorFast reduce_labels bug." + }, + { + "left": "issue:43450", + "right": "issue:43526", + "accept": false, + "reason": "Video batch-shape bug and BeitImageProcessorFast label reduction bug are different processor paths." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they involve different symbols and different root causes." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The summary matches the packet: most items are clearly unrelated by title, and the only clearly supported duplicate pair in the provided data is issue:45310 / issue:45478, which share the same Qwen3.5 Moe from_pretrained error. The soft-edge rejections are appropriately cautious, and the one accepted pair is well supported by identical titles and chronology." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 21, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:40444", + "issue:41720", + "issue:43066", + "issue:43278", + "issue:43329", + "issue:43335", + "issue:43404", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43701", + "issue:43720", + "issue:43723", + "issue:43873", + "issue:44112", + "issue:44292", + "issue:44448", + "issue:44462", + "issue:44493", + "issue:44554", + "issue:44568", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:44977", + "issue:44987", + "issue:45072", + "issue:45083", + "issue:45200", + "issue:45229", + "issue:45245", + "issue:45372", + "issue:45381", + "issue:45406", + "issue:45491", + "issue:45520", + "issue:45563" + ], + "soft_pairs": [ + "issue:44855|issue:45083", + "issue:45200|issue:45381", + "issue:39692|issue:44493", + "issue:43404|issue:45200", + "issue:43335|issue:43653", + "issue:29127|issue:32090", + "issue:44448|issue:44779", + "issue:44977|issue:45229", + "issue:38175|issue:44493", + "issue:44292|issue:45520", + "issue:43335|issue:43504", + "issue:43335|issue:45083", + "issue:43550|issue:44112", + "issue:43335|issue:45072", + "issue:44568|issue:45491", + "issue:43720|issue:44112", + "issue:44448|issue:44977", + "issue:40444|issue:43526", + "issue:43335|issue:44112", + "issue:43723|issue:44554", + "issue:34689|issue:35141", + "issue:44857|issue:45072", + "issue:43606|issue:44857", + "issue:44977|issue:45563", + "issue:45372|issue:45406", + "issue:43066|issue:44462", + "issue:43873|issue:44493", + "issue:43873|issue:45229", + "issue:43278|issue:43329", + "issue:43701|issue:45083", + "issue:29942|issue:35141", + "issue:43550|issue:45083", + "issue:41720|issue:45229", + "issue:45245|issue:45563", + "issue:32090|issue:33357", + "issue:37428|issue:44987" + ], + "prepared_review_unit_hash": "a8b3e0fece76bda5f1f91c56d74a3d231e8cc75193a7ebf373ef5088e0b01235", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 38743, + "estimated_input_tokens": 9686, + "estimated_eval_tokens": 19628 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a bag of unrelated Transformer issues across different models and subsystems. I found no convincing duplicate pairs to merge; the only reasonably strong standalone triage target is issue 45229 because it is open, high-activity, and has a concrete reproduction with broad user impact.", + "confidence": 0.84, + "canonical_issue_reason": "No clear duplicate cluster emerged; if one issue must represent the set, issue 45229 is the strongest standalone bug report thanks to detailed discussion and impact.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45229 is the most actionable and representative issue in the set: open, heavily discussed, and framed around a concrete runtime failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Different failures: Python 3.13 import/parsing error in DebertaV2 vs a Qwen3 omni feature-length helper bug." + }, + { + "left": "issue:45200", + "right": "issue:45381", + "accept": false, + "reason": "Both are multimodal, but one is about missing mm_token_type_ids defaults and the other is a Qwen2.5-VL video vision_position_ids bug." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "Doc/example failures for SigLIP2 are not the same as a generic position-id key warning across many models." + }, + { + "left": "issue:43404", + "right": "issue:45200", + "accept": false, + "reason": "Mistral3 lm_head tying and Gemma 4 mm_token_type_ids are unrelated code paths." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug is unrelated to BigBirdTokenizer special-token registration." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "An error-message clarity issue in LayoutLMv3 is not the same as a Trainer broadcast TypeError." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Both involve v5 behavior changes, but one is Pegasus output drift and the other is Deepseek tokenizer correctness; different model/code paths." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3.5 flash-attention generation failure and Gemma4 multi-GPU CUDA OOM are different symptoms with different root causes." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 is not the same bug as widespread unexpected position-id keys." + }, + { + "left": "issue:44292", + "right": "issue:45520", + "accept": false, + "reason": "Qwen-3-8B-NVFP4 runtime failure and a Python 3.13 flash_attn import KeyError are unrelated." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Sparse-layer creation in SwitchTransformers and BEiT legacy-field loading are different bugs." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "A SwitchTransformers config edge case does not match a Qwen3 Omni feature-length helper regression." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba is unrelated to a stale GraniteSpeech CI test override." + }, + { + "left": "issue:43335", + "right": "issue:45072", + "accept": false, + "reason": "SwitchTransformers config logic and bfloat16 dtype mismatches in inference are not the same underlying issue." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer special-token handling for mdeberta-v3-base is unrelated to Gemma3 NaN embeddings from sliding-window attention." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "BitNet packed-weight loading and a stale CI test are completely different concerns." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Pegasus v4/v5 output differences and Qwen3.5 flash-attention generation issues are distinct model regressions." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Qwen2.5-VL iterable-dataset multi-image finetuning failure is unrelated to BEiT fast processor label reduction." + }, + { + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "SwitchTransformers sparse-layer config and GraniteSpeech device-override CI failure do not share a root bug." + }, + { + "left": "issue:43723", + "right": "issue:44554", + "accept": false, + "reason": "Tokenizer loading in v5 and an MPS attention correctness issue are different subsystems and failures." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Llama 3.2 Vision model loading regression and embedding retie after resize are not the same bug." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Float16 AMP crash in LwDetrImageLoss is unrelated to bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "CPU offload device mismatch and an AMP/CUDA float16 crash are separate runtime issues." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Flash-attention generation failure and a stale warning about num_return_sequences are not the same problem." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Both involve Gemma4, but one is a mistral_common import error during processor loading and the other is a missing _tokenizer attribute in serve." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Wrong tokenizer decoder type and ignoring tokenizer.json are related broadly to tokenizers, but they target different behaviors and fixes." + }, + { + "left": "issue:43873", + "right": "issue:44493", + "accept": false, + "reason": "Quantization/offloading behavior and unexpected position-id keys are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Offloading with quantization and Gemma4 multi-GPU CUDA OOM are different resource-management failures." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "Embedding dtype drift between train/eval is unrelated to the multimodal token-count video-branch bug." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "resume_from_checkpoint key mismatch does not match the Qwen3 Omni feature-length helper issue." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention 2 test failures are unrelated to embedding reinitialization after token resize." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure is not the same as a Qwen3 Omni helper bug." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3 auto device-mapping assert and Gemma4 multi-GPU OOM are different model/runtime failures." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-count limit error and a stale paged-generate warning are unrelated." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Trainer NoneType broadcast failure and a MacOS CLIP bus error are different issues." + }, + { + "left": "issue:37428", + "right": "issue:44987", + "accept": false, + "reason": "Flash-attention import-name failure in Python 3.13 and a transformers>=5.1.0 model-loading failure are different root causes." + } + ] + }, + "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": 13, + "review_unit_index": 22, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:25251", + "issue:29127", + "issue:29942", + "issue:30064", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34567", + "issue:39692", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43257", + "issue:43278", + "issue:43381", + "issue:43421", + "issue:43504", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43824", + "issue:43976", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44279", + "issue:44361", + "issue:44442", + "issue:44448", + "issue:44451", + "issue:44466", + "issue:44479", + "issue:44610", + "issue:44743", + "issue:44857", + "issue:44871", + "issue:44977", + "issue:45072", + "issue:45103", + "issue:45198", + "issue:45229", + "issue:45372", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45520" + ], + "soft_pairs": [ + "issue:43976|issue:45440", + "issue:45372|issue:45459", + "issue:44448|issue:44466", + "issue:44451|issue:45520", + "issue:25251|issue:32090", + "issue:43065|issue:45412", + "issue:43065|issue:43994", + "issue:44186|issue:45198", + "issue:43278|issue:43381", + "issue:43653|issue:44743", + "issue:44265|issue:45198", + "issue:44361|issue:45198", + "issue:43257|issue:43278", + "issue:44871|issue:44977", + "issue:43421|issue:45491", + "issue:43994|issue:45412", + "issue:42915|issue:42994", + "issue:44206|issue:44479", + "issue:43756|issue:45440", + "issue:43504|issue:44857", + "issue:29942|issue:30064", + "issue:44610|issue:45412", + "issue:33357|issue:39692", + "issue:32090|issue:33290", + "issue:45103|issue:45372", + "issue:29127|issue:34567", + "issue:45072|issue:45229", + "issue:43824|issue:44279", + "issue:43381|issue:43701", + "issue:43653|issue:44442" + ], + "prepared_review_unit_hash": "7bedd4ceaa3381622f7c6ec11657994b95725c2893c40df267074164ef6bdfaf", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36666, + "estimated_input_tokens": 9167, + "estimated_eval_tokens": 18590 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous similarity bucket, not a duplicate cluster. The suggested pairs share broad library/model keywords but describe different bugs, regressions, or features, so none should be merged as the same underlying issue.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items are unrelated enough that none is a defensible duplicate anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a good global representative for this cluster; the issues span distinct subsystems and failure modes.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python-version compatibility issue vs DeepSeekV3 implementation divergence; different code paths and failures." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma 4 processor import failure vs tokenizer error masking when protobuf is missing; not the same bug." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Different problems: Pegasus output regression vs inconsistent lm_head serialization." + }, + { + "left": "issue:44451", + "right": "issue:45520", + "accept": false, + "reason": "Model loading failure for a specific checkpoint vs flash_attn import KeyError on Python 3.13." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting bug is unrelated to Trainer _gpu_broadcast_one NoneType error." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder vs RT-DETR memory not being released; different defects." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "SAM3 pixel decoder placeholder layer vs SigLIP2 AutoModel/pipeline misbehavior." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2 tokenizer crash on NER/padding vs Wav2Vec2 save_pretrained/tokenization failure." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Embedding dtype drift in evaluation vs gradient checkpointing being unsupported in eval mode." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBird special-token registration bug vs Qwen3.5 recurrent-state reset bug." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export/torch_compilable_check failure vs Wav2Vec2 save/tokenization failure." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer AttributeError vs Wav2Vec2 save/tokenization failure; unrelated components." + }, + { + "left": "issue:43257", + "right": "issue:43278", + "accept": false, + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed vs embedding dtype mismatch during eval." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma eos_token_id config inconsistency vs Qwen3.5 flash-attention generation regression." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime post-processor update bug in TokenizersBackend vs NaN embeddings from Gemma3 sliding-window attention." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 inference/output issue vs RT-DETR memory leak; no shared failing code path." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "FineGrainedFP8Config failure vs quantized model saving failure are different quantization problems." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Unsupported center argument in LasrFeatureExtractor vs Qwen VL video-input regression." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch vs DeepSeekV3 implementation divergence; different architectures and symptoms." + }, + { + "left": "issue:43504", + "right": "issue:44857", + "accept": false, + "reason": "Legacy field prevents BEiT preset loading vs LwDetrImageLoss AMP/CUDA crash." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures vs void segmentation-map processing bug." + }, + { + "left": "issue:44610", + "right": "issue:45412", + "accept": false, + "reason": "Processor/model input-size mismatch vs RT-DETR garbage-collection memory issue." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "MacOS bus error on CLIP model vs SigLIP2 documentation example errors; distinct models and failures." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Trainer broadcast NoneType error vs Adafactor+DeepSpeed OOM; unrelated causes." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotations crash vs Gemma 4 processor import dependency issue." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error-message clarity vs TrainerState token-count tracking bug." + }, + { + "left": "issue:45072", + "right": "issue:45229", + "accept": false, + "reason": "bfloat16 dtype mismatches in inference vs Gemma4 multi-GPU CUDA OOM." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "Missing Qwen2_5_VL import vs generic transformers dependency issue; not the same bug." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode vs resume_from_checkpoint key mismatch." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBird special-token registration bug vs AutoTokenizer loading failure for FastSpeech2ConformerTokenizer." + } + ] + }, + "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": 13, + "review_unit_index": 23, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:33357", + "issue:33666", + "issue:35141", + "issue:36010", + "issue:39692", + "issue:42915", + "issue:43232", + "issue:43262", + "issue:43316", + "issue:43334", + "issue:43388", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43575", + "issue:43646", + "issue:43701", + "issue:43723", + "issue:43761", + "issue:43931", + "issue:43994", + "issue:44077", + "issue:44079", + "issue:44186", + "issue:44188", + "issue:44206", + "issue:44387", + "issue:44442", + "issue:44464", + "issue:44479", + "issue:44610", + "issue:44743", + "issue:44811", + "issue:44857", + "issue:44898", + "issue:44964", + "issue:44977", + "issue:45003", + "issue:45072", + "issue:45216", + "issue:45245", + "issue:45290", + "issue:45440", + "issue:45479", + "issue:45491" + ], + "soft_pairs": [ + "issue:44464|issue:45491", + "issue:43526|issue:44186", + "issue:43334|issue:43531", + "issue:43262|issue:44206", + "issue:29942|issue:36010", + "issue:44977|issue:45245", + "issue:29127|issue:33357", + "issue:29127|issue:35141", + "issue:43761|issue:44079", + "issue:44442|issue:45072", + "issue:42915|issue:43931", + "issue:43232|issue:43388", + "issue:43526|issue:44857", + "issue:43526|issue:44610", + "issue:44743|issue:44811", + "issue:44479|issue:45290", + "issue:43723|issue:45479", + "issue:45003|issue:45440", + "issue:43575|issue:43701", + "issue:43931|issue:45216", + "issue:43646|issue:44077", + "issue:43761|issue:44188", + "issue:43575|issue:44387", + "issue:44964|issue:45440", + "issue:29942|issue:33666", + "issue:43994|issue:44206", + "issue:43316|issue:43525", + "issue:39692|issue:45491", + "issue:44442|issue:44898" + ], + "prepared_review_unit_hash": "d836530ed7e88d8f737171749d62132d7285e72f7cd53635d34fa71e0b2585ce", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35919, + "estimated_input_tokens": 8980, + "estimated_eval_tokens": 18216 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is highly heterogeneous: most items are unrelated regressions across different models, processors, and APIs. I found no soft pair that clearly represents the same underlying bug/change, so I would not merge any of the candidate edges.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 43646 is the broadest version-regression item in the set and the best single umbrella anchor if one must be chosen, but the overall cluster is too mixed for a true canonical duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "43646 is the most general, high-level issue title among otherwise unrelated reports, so it is the safest representative issue for triage, though not a real duplicate hub.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44464", + "right": "issue:45491", + "accept": false, + "reason": "Both involve generation/attention behavior, but one is chunked generation with compiled forward and the other is Gemma3 NaN embeddings on mixed-length batches; different code paths and failures." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Both are preprocessing/label-handling bugs, but they affect different components and symptoms (BeitImageProcessorFast label reduction vs LayoutLMv2Tokenizer NER/padding crashes)." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Both are Qwen3-family issues, but one is a missing `pad_token_id` on load and the other is a `sliding_window` behavior bug in Qwen3-MoE; not the same defect." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Both touch audio processing, but one is a chat-template sampling-rate default and the other is an unsupported `center` argument in a feature extractor; distinct bugs." + }, + { + "left": "issue:29942", + "right": "issue:36010", + "accept": false, + "reason": "FlashAttention test failures and a `GenerationMixin` import error are unrelated failure modes with no shared concrete code path." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3.5 flash-attention generation failure and the 2^24 category limit error are entirely different problems." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 error messaging and a MacOS bus error in a CLIP community model are unrelated." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "Different subsystems: LayoutLMv3 box-info validation vs embedding resize/post-init reinitialization." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "One is a CLIPVisionModel hidden-states regression; the other is a generic `ModelOutput` key-assignment issue. They may be loosely related, but not the same concrete bug." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "AutoTokenizer loading failure and bfloat16 dtype mismatches are different classes of bugs." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "Both mention Qwen3 models, but one is FP8 quantization failure and the other is a weight-shape mismatch on load; no clear shared code path." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs syncing and metric-gathering label truncation are unrelated training/runtime paths." + }, + { + "left": "issue:43526", + "right": "issue:44857", + "accept": false, + "reason": "A label reduction bug in BeitImageProcessorFast is not the same as an AMP/CUDA crash in LwDetrImageLoss." + }, + { + "left": "issue:43526", + "right": "issue:44610", + "accept": false, + "reason": "Both concern image preprocessing, but one is label reduction and the other is incorrect output resolution; different underlying bugs." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Qwen3.5 recurrent-state reset and Whisper `batch_decode` skipping special tokens are unrelated issues." + }, + { + "left": "issue:44479", + "right": "issue:45290", + "accept": false, + "reason": "One is a video-input regression across Qwen VL models; the other is a chat-template crash on assistant tool-call messages. Different features and code paths." + }, + { + "left": "issue:43723", + "right": "issue:45479", + "accept": false, + "reason": "Tokenizer loading in v5 and single-label classification zero-loss are unrelated." + }, + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "Unsafe `sys.modules` access in modeling_utils and a DeepSeekV3MoE divergence are separate implementation areas." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Tensor-parallel OOM on model load and resume-from-checkpoint key mismatch are not the same issue." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Both are Qwen3.5/Qwen3-VL loading/saving issues, but one is a checkpoint shape mismatch and the other is an incorrect `save_pretrained` checkpoint; related family, different bug." + }, + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Generic custom-model init breakage and the `patchtsmixer` post_init policy issue do not share the same failure mechanism." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "A CLIPVisionModel output regression and a torch.compile attention-kernel divergence are different code paths." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "OOM from tensor-parallel loading and increased reserved memory under int4 quantization are both memory issues, but not the same root cause." + }, + { + "left": "issue:44964", + "right": "issue:45440", + "accept": false, + "reason": "Phi-4 multimodal load failure and DeepSeekV3MoE divergence are unrelated model-specific issues." + }, + { + "left": "issue:29942", + "right": "issue:33666", + "accept": false, + "reason": "Flash Attention 2 test failures and Qwen2-VL multi-GPU training are different problem domains." + }, + { + "left": "issue:43994", + "right": "issue:44206", + "accept": false, + "reason": "SigLIP2 model/pipeline nonsense results and an audio feature extractor argument crash are unrelated." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Both mention config API discrepancies, but they affect different configs and different missing attributes; too broad to be a duplicate." + }, + { + "left": "issue:39692", + "right": "issue:45491", + "accept": false, + "reason": "SigLIP2 doc-example problems and Gemma3 NaN embeddings are unrelated." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "AutoTokenizer loading failure and Perceiver non-default-resolution failure are different bugs in different components." + } + ] + }, + "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": 13, + "review_unit_index": 24, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:25251", + "issue:28282", + "issue:33290", + "issue:34567", + "issue:34689", + "issue:35141", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41628", + "issue:42175", + "issue:43065", + "issue:43316", + "issue:43381", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43493", + "issue:43540", + "issue:43611", + "issue:43643", + "issue:43673", + "issue:43688", + "issue:43720", + "issue:43749", + "issue:43827", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:44077", + "issue:44162", + "issue:44190", + "issue:44206", + "issue:44442", + "issue:44479", + "issue:44496", + "issue:44514", + "issue:44617", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44871", + "issue:44877", + "issue:44913", + "issue:45137", + "issue:45372" + ], + "soft_pairs": [ + "issue:33290|issue:34567", + "issue:43673|issue:44743", + "issue:34689|issue:43611", + "issue:43381|issue:43873", + "issue:43493|issue:43827", + "issue:33290|issue:35141", + "issue:33290|issue:34689", + "issue:44077|issue:44479", + "issue:43720|issue:44442", + "issue:41628|issue:45372", + "issue:43749|issue:43931", + "issue:43643|issue:44913", + "issue:43873|issue:44871", + "issue:39692|issue:44206", + "issue:44190|issue:44496", + "issue:43381|issue:43856", + "issue:33290|issue:36331", + "issue:34689|issue:36010", + "issue:44855|issue:45137", + "issue:43540|issue:44514", + "issue:43316|issue:44877", + "issue:44162|issue:44792", + "issue:43388|issue:43688", + "issue:42175|issue:43316", + "issue:16998|issue:22355", + "issue:43404|issue:43441", + "issue:25251|issue:28282", + "issue:43065|issue:44617" + ], + "prepared_review_unit_hash": "74dc5a548a9309b5d734550bac2e433f131f9fe0ecd938b5b5e7bafaf8d8887e", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36072, + "estimated_input_tokens": 9018, + "estimated_eval_tokens": 18292 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated Transformers issues with overlapping vocabulary. None of the soft pairs look like true duplicates or mergeable equivalents of the same underlying bug/change.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits this cluster: the items span unrelated imports, configs, training bugs, model-loading regressions, and docs problems, so there is no shared root cause to dedupe around.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a good global representative; the similarity here is mostly lexical, not causal, and the problems target different code paths and models.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:33290", + "right": "issue:34567", + "accept": false, + "reason": "Optimizer OOM in DeepSpeed and TrainerState token-count update are different training subsystems and bugs." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Missing GenerationMixin cache and Qwen3.5 recurrent-state reset are different generation/cache issues." + }, + { + "left": "issue:34689", + "right": "issue:43611", + "accept": false, + "reason": "Both are loading regressions, but they affect different mechanisms and models: Llama 3.2 vision loading vs base_model_prefix handling." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Gradient checkpointing in eval mode and quantized offloading are unrelated behaviors with no shared code path." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and docs still using pipeline() are different bug classes." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "AdaFactor/DeepSpeed OOM and embedding reinitialization after resize are unrelated training issues." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Optimizer memory usage and model-loading regression are different problems in different parts of the stack." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "Optional post_init on patchtsmixer and video-input regression for Qwen variants are not the same bug." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "BitNet packed-weight loading and FastSpeech2Conformer tokenizer import/load failure are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:45372", + "accept": false, + "reason": "Both involve imports, but they are different missing symbols from different optional dependencies and processor paths." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU-efficient loading and Qwen3-VL weight-shape mismatch are distinct loading failures." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "Missing fields with trust_remote_code and GPTNeoX rotary_pct resetting on reload are separate config bugs." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization/offloading behavior and Gemma-3 eos_token_id inconsistency are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:44206", + "accept": false, + "reason": "SigLIP2 docs/example problems and LasrFeatureExtractor center-arg regression are different issues." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Local dataset loading in an example script and unrecognized model/config loading are unrelated." + }, + { + "left": "issue:43381", + "right": "issue:43856", + "accept": false, + "reason": "Checkpointing-in-eval and Qwen3 MoE memory usage are different training/runtime issues." + }, + { + "left": "issue:33290", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed OOM and CustomTrainer compute_loss signature breakage are not the same code-path problem." + }, + { + "left": "issue:34689", + "right": "issue:36010", + "accept": false, + "reason": "Llama 3.2 vision loading regression and GenerationMixin import-path break are distinct import/load regressions." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 IndentationError in DebertaV2Model and ZeRO3 deque pop error are unrelated failures." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both touch multimodal inputs, but Qwen3Omni video processing and Qwen2_5_VL batched chat-template padding failure are different bugs." + }, + { + "left": "issue:43316", + "right": "issue:44877", + "accept": false, + "reason": "Gemma3TextConfig API mismatch and granite_speech strict-config loading are separate config issues." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 breakage and janus image-generation test failure are unrelated model-specific issues." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Label dropping in gather_for_metrics and auxiliary-loss normalization are different training/evaluation code paths." + }, + { + "left": "issue:42175", + "right": "issue:43316", + "accept": false, + "reason": "Packaging backend omission and Gemma3TextConfig API discrepancy are unrelated." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "Model_max_length question for DeBERTa-V3 and missing transformers.onnx import are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Mistral3 lm_head tying and Ministral-3 FlashAttention failure are different model-specific bugs." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline top_k output shape change and missing PyTorch import are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "Sam3PixelDecoder dummy Conv2d and Sam3Video CUDA OOM are different issues; one is architecture scaffolding, the other memory pressure." + } + ] + }, + "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": 13, + "review_unit_index": 25, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:32090", + "issue:35141", + "issue:41950", + "issue:42617", + "issue:43065", + "issue:43295", + "issue:43316", + "issue:43452", + "issue:43475", + "issue:43525", + "issue:43531", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43701", + "issue:43716", + "issue:43742", + "issue:43761", + "issue:43825", + "issue:43827", + "issue:43873", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44060", + "issue:44077", + "issue:44242", + "issue:44464", + "issue:44479", + "issue:44496", + "issue:44521", + "issue:44534", + "issue:44610", + "issue:44734", + "issue:44792", + "issue:44871", + "issue:44913", + "issue:44933", + "issue:44977", + "issue:45127", + "issue:45200", + "issue:45290", + "issue:45362", + "issue:45381", + "issue:45406", + "issue:45563" + ], + "soft_pairs": [ + "issue:43316|issue:43716", + "issue:43295|issue:43825", + "issue:41950|issue:43475", + "issue:45127|issue:45200", + "issue:43761|issue:44479", + "issue:44077|issue:44534", + "issue:43550|issue:43701", + "issue:43295|issue:44871", + "issue:43606|issue:43701", + "issue:44734|issue:45406", + "issue:43646|issue:44913", + "issue:43452|issue:44496", + "issue:43906|issue:45290", + "issue:43575|issue:43742", + "issue:43761|issue:44242", + "issue:42617|issue:43575", + "issue:43931|issue:44060", + "issue:44521|issue:45381", + "issue:43575|issue:43873", + "issue:32090|issue:35141", + "issue:44464|issue:44871", + "issue:44464|issue:45563", + "issue:44479|issue:44734", + "issue:43525|issue:43531", + "issue:44792|issue:44977", + "issue:44933|issue:45362", + "issue:43827|issue:43994", + "issue:43065|issue:44610", + "issue:44534|issue:44913" + ], + "prepared_review_unit_hash": "689ee891aaf0fabad24ab24e55c21b4e56fae00df273e09a015d9a29af1a2c79", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35697, + "estimated_input_tokens": 8925, + "estimated_eval_tokens": 18106 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the items span unrelated bugs in model loading, chat templates, multimodal processors, generation, config serialization, offloading, and serving. I did not find any soft pair that clearly refers to the same underlying bug/change.", + "confidence": 0.94, + "canonical_issue_reason": "No strong canonical issue exists because the cluster is not a single duplicate family. If one issue must be chosen, issue 44521 is the broadest user-facing regression and has the most cross-references.", + "canonical_pr_reason": null, + "best_issue_reason": "issue 44521 is the most broadly impactful and actively referenced item in the set, with multiple inbound references and a cross-cutting multimodal template regression.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "Different subsystems and symptoms: Gemma3 config API mismatch vs Mistral-3 image preprocessor dtype mismatch." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Both are v5 regressions, but one is processor/tokenizer access and image passing; the other is a pipeline error message about translation support." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Both involve video/image models, but the concrete failures differ: wrong processor lookup vs missing fpn_position_embeddings attribute." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "Unrelated bugs: LoRA merge with extended vocab causes collapse, while Gemma 4 text-only fine-tuning needs default mm_token_type_ids." + }, + { + "left": "issue:43761", + "right": "issue:44479", + "accept": false, + "reason": "Different model families and failure modes: CLIPVision hidden_states regression vs Qwen VL video-input regression." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "One is about optional post_init in patchtsmixer; the other is a v5 non-persistent buffer initialization problem." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "Torch.compile SDPA failure for Bamba is unrelated to resume_from_checkpoint key mismatches." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Processor/tokenizer regression is unrelated to Gemma-3 eos_token_id inconsistency." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch is not the same bug as checkpoint resume key mismatch." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Both are serve-related, but one is KV-cache tensor indexing and the other is missing _tokenizer on Gemma4Processor." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Generic custom model init breakage is not the same as GPTNeoX rotary_pct not persisting on reload." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "Both are model-loading failures, but one is gguf_file/from_pretrained behavior and the other is a missing model_type in config.json." + }, + { + "left": "issue:43906", + "right": "issue:45290", + "accept": false, + "reason": "Different chat/template errors: isolated reproduction of an older issue vs assistant messages with tool calls and no content." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "OOM on large Qwen2 model with tensor parallelism is unrelated to a key error loading MobileLLM-125M." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "Hidden_states regression in CLIPVision is unrelated to router-logits/load-balancing loss behavior." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "3d_parallel.py execution failure is not the same concrete issue as Qwen2 loading OOM under tp." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Weight-shape mismatch loading Qwen3-VL is distinct from the incorrect tied-weights warning in Qwen3-Next." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "Both are multimodal regressions, but one is assistant mask generation and the other is video vision_position_ids." + }, + { + "left": "issue:43575", + "right": "issue:43873", + "accept": false, + "reason": "Memory OOM under tensor parallelism is not the same as quantization/offloading not behaving as expected." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Trainer broadcast TypeError and post_init embedding reinitialization are unrelated bugs." + }, + { + "left": "issue:44464", + "right": "issue:44871", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency is unrelated to Gemma-3 eos_token_id configuration drift." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "A generation correctness bug is not the same as a stale warning about num_return_sequences." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Video input regression in Qwen VL is unrelated to KV-cache continuation indexing in serve." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config is unrelated to Qwen3-MoE sliding_window behavior." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Janus image-generation test failure does not match Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Import error from image_utils is unrelated to Qwen3.5 chat crashes." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs still referencing pipeline() is not the same as SigLIP2 producing bad outputs with AutoModel/pipeline." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder is unrelated to OmDet-Turbo processor/model resolution mismatch." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffer junk in v5 is a different failure mode from rotary_pct not surviving reload." + } + ] + }, + "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": 13, + "review_unit_index": 26, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36246", + "issue:38175", + "issue:38617", + "issue:39692", + "issue:41628", + "issue:41720", + "issue:41950", + "issue:43065", + "issue:43116", + "issue:43295", + "issue:43335", + "issue:43388", + "issue:43404", + "issue:43479", + "issue:43540", + "issue:43550", + "issue:43582", + "issue:43644", + "issue:43650", + "issue:43761", + "issue:43976", + "issue:44016", + "issue:44077", + "issue:44079", + "issue:44186", + "issue:44190", + "issue:44220", + "issue:44246", + "issue:44355", + "issue:44466", + "issue:44496", + "issue:44734", + "issue:44743", + "issue:44855", + "issue:44871", + "issue:44913", + "issue:44938", + "issue:44964", + "issue:45003", + "issue:45127", + "issue:45216", + "issue:45245", + "issue:45276", + "issue:45357", + "issue:45372", + "issue:45375", + "issue:45542" + ], + "soft_pairs": [ + "issue:41720|issue:45245", + "issue:44938|issue:45542", + "issue:44190|issue:44964", + "issue:43388|issue:43582", + "issue:43650|issue:44220", + "issue:45127|issue:45357", + "issue:43761|issue:44743", + "issue:43644|issue:44913", + "issue:44077|issue:44079", + "issue:44355|issue:44855", + "issue:43976|issue:44855", + "issue:43540|issue:44734", + "issue:44496|issue:45372", + "issue:43295|issue:45276", + "issue:43550|issue:44190", + "issue:43335|issue:44855", + "issue:43404|issue:45216", + "issue:43116|issue:44190", + "issue:39692|issue:44190", + "issue:43644|issue:44871", + "issue:41950|issue:44016", + "issue:43479|issue:45375", + "issue:36246|issue:41628", + "issue:44186|issue:44734", + "issue:44466|issue:44913", + "issue:38175|issue:43065", + "issue:36246|issue:38617", + "issue:44246|issue:45003" + ], + "prepared_review_unit_hash": "896ab4dbeee29e631fce75c5df075bf4efd799d818849c0171233dad0694d671", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35881, + "estimated_input_tokens": 8971, + "estimated_eval_tokens": 18198 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous and does not form one real duplicate cluster. The only close duplicate-like thread is the Qwen3.5 save_pretrained regression (issue 45216, with follow-up 45357); all proposed soft edges are false positives.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 45216 is the best canonical among the set because it is the broadest, earliest report of the Qwen3.5 save_pretrained regression, and issue 45357 reads like a later follow-up on the same checkpoint/visual-key corruption.", + "canonical_pr_reason": null, + "best_issue_reason": "45216 is the clearest representative issue: concrete regression, explicit version boundary, and strong overlap with the later Qwen3.5 save_pretrained report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Different failures in unrelated areas: Qwen3 auto device mapping CUDA assert vs category-count limit error." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python 3.14 import failure is unrelated to the TensorFlow/tensorboard backend issue." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Different models and symptoms: local dataset loading for an example script vs loading Phi-4 multimodal." + }, + { + "left": "issue:43388", + "right": "issue:43582", + "accept": false, + "reason": "Training metric batching bug is unrelated to Apple Silicon allocator warmup TypeError." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "No clear shared bug; one is a placeholder/spam-like issue, the other is an audio feature extraction problem." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen-related, but the bugs differ: LoRA/extended-vocab collapse vs save_pretrained visual-key serialization." + }, + { + "left": "issue:43761", + "right": "issue:44743", + "accept": false, + "reason": "Different model families and code paths: CLIPVision hidden_states vs Qwen recurrent cache reset." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Generic v5 buffer serialization bug vs GPTNeoXConfig rotary_pct reload regression are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44079", + "accept": false, + "reason": "Separate core bugs: patchtsmixer post_init validation vs ModelOutput key assignment." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Compiled Python-file errors are not the same as the DebertaV2 Python 3.13 IndentationError." + }, + { + "left": "issue:43976", + "right": "issue:44855", + "accept": false, + "reason": "Version compatibility failure is unrelated to the decorator/comment parsing bug in DebertaV2." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Video input processing in Qwen3OmniMoe is unrelated to serve KV-cache tensor indexing." + }, + { + "left": "issue:44496", + "right": "issue:45372", + "accept": false, + "reason": "Unrecognized model/config error is unrelated to the mistral_common ReasoningEffort import breakage." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Processor/tokenizer regression is unrelated to resize_token_embeddings not updating Gemma4 embeddings." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "torch.compile/SDPA failure is unrelated to the image-classification no-trainer dataset loading problem." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug is unrelated to the Python 3.13 parsing/IndentationError." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Both involve tied weights/checkpointing, but they are different model families and different concrete bugs." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Empty multi-label classification outputs are unrelated to local dataset loading in an example script." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "SigLIP2 documentation/model mismatch is unrelated to the no-trainer image-classification dataset issue." + }, + { + "left": "issue:43644", + "right": "issue:44871", + "accept": false, + "reason": "Non-persistent buffer junk in v5 is unrelated to Gemma-3 eos_token_id config mismatch." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Video-classification processor lookup bug is unrelated to a notebook syntax error." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Different config bugs in different multimodal models; not the same underlying issue." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but for different missing symbols and different API surfaces." + }, + { + "left": "issue:44186", + "right": "issue:44734", + "accept": false, + "reason": "LayoutLMv2 tokenizer NER/padding crash is unrelated to serve KV-cache continuation indexing." + }, + { + "left": "issue:44466", + "right": "issue:44913", + "accept": false, + "reason": "Inconsistent lm_head.weight serialization is unrelated to GPTNeoX rotary_pct reload behavior." + }, + { + "left": "issue:38175", + "right": "issue:43065", + "accept": false, + "reason": "SigLIP2 zero probabilities and Sam3PixelDecoder dummy Conv2d are different model-specific bugs." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Different import failures: missing Qwen2_5_VLImageProcessor vs missing layer_type_validation." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Import slowness is a performance complaint, not the sys.modules safety bug." + } + ] + }, + "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": 13, + "review_unit_index": 27, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:41950", + "issue:42175", + "issue:42907", + "issue:43295", + "issue:43299", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43526", + "issue:43577", + "issue:43643", + "issue:43650", + "issue:43673", + "issue:43824", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44075", + "issue:44112", + "issue:44220", + "issue:44242", + "issue:44279", + "issue:44336", + "issue:44351", + "issue:44368", + "issue:44496", + "issue:44521", + "issue:44683", + "issue:44792", + "issue:44871", + "issue:44933", + "issue:44964", + "issue:44987", + "issue:45042", + "issue:45200", + "issue:45216", + "issue:45245", + "issue:45278", + "issue:45325", + "issue:45335", + "issue:45357", + "issue:45405", + "issue:45446", + "issue:45542", + "issue:45561", + "issue:45563" + ], + "soft_pairs": [ + "issue:44683|issue:45446", + "issue:44016|issue:44220", + "issue:42175|issue:44279", + "issue:44112|issue:44792", + "issue:43295|issue:45335", + "issue:43643|issue:44496", + "issue:45561|issue:45563", + "issue:43299|issue:45216", + "issue:42907|issue:43441", + "issue:44016|issue:44062", + "issue:44112|issue:44964", + "issue:44871|issue:45245", + "issue:43673|issue:44336", + "issue:43854|issue:44964", + "issue:43976|issue:44987", + "issue:44496|issue:45042", + "issue:43526|issue:43901", + "issue:43824|issue:44933", + "issue:43976|issue:44279", + "issue:43299|issue:45357", + "issue:43577|issue:45216", + "issue:44075|issue:44368", + "issue:44521|issue:45325", + "issue:43901|issue:44496", + "issue:44242|issue:45200", + "issue:44683|issue:45542", + "issue:43976|issue:45405", + "issue:44351|issue:45278", + "issue:43479|issue:43525", + "issue:41950|issue:43650" + ], + "prepared_review_unit_hash": "d4fd8bd9aee055224ac3b07d063722a2abfb5f6e01c68afe195e57fd75d6db0a", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36583, + "estimated_input_tokens": 9146, + "estimated_eval_tokens": 18548 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous Transformers issues with only superficial similarity between a few pairs. I would not merge any of the soft-edge candidates; they share broad subsystems or model families, but not the same concrete bug/change.", + "confidence": 0.46, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella-style report in the set, covering the import-regression theme seen in several nearby issues, but the cluster is still too mixed for a perfect canonical issue.", + "canonical_pr_reason": null, + "best_issue_reason": "45278 is the best representative issue because it captures the widest shared symptom class (post-upgrade import breakage) among the otherwise scattered reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Both touch flex_attention, but one is a torch>=2.9 compilation failure and the other is a version-check bug for AuxRequest import; different root causes." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "A notebook syntax error and a feature-extraction function issue are unrelated bugs." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "One is a missing TensorFlow backend in the torch extra, the other is a generic dependency complaint; not the same concrete defect." + }, + { + "left": "issue:44112", + "right": "issue:44792", + "accept": false, + "reason": "Different models and different failure modes: CI stale test in GraniteSpeech vs janus image-generation test failure." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Both are regressions, but one is about processor/tokenizer image handling and the other about resize_token_embeddings not updating decoder.embed_tokens." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "Both involve config loading, but one is missing fields with trust_remote_code and the other is an unrecognized model_type error." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "A pytest-xdist file race and a stale generation warning are unrelated issues." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Same model family, but loading a Qwen3.5 MoE model and save_pretrained saving wrong keys are distinct code paths." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Different Ministral failures: dequantized save path vs FlashAttention inference failure." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error vs AddedToken keyword-argument error; no shared underlying bug." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "CI test flake in GraniteSpeech is unrelated to Phi-4 multimodal loading." + }, + { + "left": "issue:44871", + "right": "issue:45245", + "accept": false, + "reason": "Gemma eos_token_id inconsistency and a category-count runtime error are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Chunked prefill cache handling and ANSI output in loading_report do not describe the same bug." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading failures, but for different models and likely different causes." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "A Python version support issue and a model-loading regression are not duplicates." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Unrecognized model config vs PIL image-processor torchvision dependency are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "A processor bug and a documentation mismatch are not the same issue." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Both are import-related, but they reference different symbols and different failure contexts." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "Python compatibility and dependency installation issues are not the same bug." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Both concern Qwen3.5 regressions, but loading failure and incorrect saved visual encoder keys are distinct problems." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "Blip2 dtype loading and Qwen3.5 checkpoint serialization are unrelated." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "Optimizer arguments being ignored and a tie_word_embeddings warning are unrelated." + }, + { + "left": "issue:44521", + "right": "issue:45325", + "accept": false, + "reason": "Both are multimodal, but assistant mask generation and rope-index scaling are separate code paths." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "Documentation about return_all_scores does not duplicate an unrecognized-model loading error." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "MoE load-balancing loss and Gemma 4 mm_token_type_ids are different training/config bugs." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "FlexAttention on torch>=2.9 and a missing TensorFlow backend are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python support and a PEFT version bump are not the same issue." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "Both mention imports, but 45278 is a broad upgrade report and 44351 is a specific HybridCache import failure; not enough evidence of one root cause." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Different config classes and different missing-default behavior; not duplicates." + }, + { + "left": "issue:41950", + "right": "issue:43650", + "accept": false, + "reason": "A pipeline processor lookup bug and a placeholder issue title 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": 2, + "component_count": 13, + "review_unit_index": 28, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:38175", + "issue:39692", + "issue:41950", + "issue:42994", + "issue:43065", + "issue:43122", + "issue:43262", + "issue:43295", + "issue:43299", + "issue:43317", + "issue:43441", + "issue:43450", + "issue:43550", + "issue:43650", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43784", + "issue:43854", + "issue:43901", + "issue:44016", + "issue:44162", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44291", + "issue:44297", + "issue:44462", + "issue:44466", + "issue:44496", + "issue:44521", + "issue:44617", + "issue:44623", + "issue:44671", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:44877", + "issue:44898", + "issue:44964", + "issue:44987", + "issue:45020", + "issue:45081", + "issue:45276", + "issue:45290", + "issue:45397", + "issue:45405", + "issue:45542" + ], + "soft_pairs": [ + "issue:44496|issue:44877", + "issue:44462|issue:45081", + "issue:41950|issue:45020", + "issue:43550|issue:44964", + "issue:44297|issue:44462", + "issue:44279|issue:45405", + "issue:43299|issue:43854", + "issue:43295|issue:44779", + "issue:44462|issue:44779", + "issue:38175|issue:44617", + "issue:39692|issue:43901", + "issue:43650|issue:45290", + "issue:44016|issue:44291", + "issue:44466|issue:45276", + "issue:43450|issue:44279", + "issue:43723|issue:44162", + "issue:43262|issue:44521", + "issue:43441|issue:44683", + "issue:43065|issue:44898", + "issue:42994|issue:43317", + "issue:44261|issue:44263", + "issue:44623|issue:45397", + "issue:43122|issue:43650", + "issue:44623|issue:44749", + "issue:43784|issue:45542", + "issue:44671|issue:44779", + "issue:43673|issue:43761", + "issue:44987|issue:45542" + ], + "prepared_review_unit_hash": "bc74b5b1033f1addd791a28eaae5eb05227f927665fc6b11c7a1e975d7afda67", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36047, + "estimated_input_tokens": 9012, + "estimated_eval_tokens": 18280 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous: mostly unrelated Transformers v5 regressions across tokenizers, chat templates, model loading, quantization, and docs. None of the soft edges look like true duplicates of the same underlying bug/change, so I reject all of them. The best representative issue is the broad remote_code loading regression.", + "confidence": 0.67, + "canonical_issue_reason": "Issue 45020 is the broadest, highest-signal representative of the set: it captures a wide class of recent-version model loading breakages around remote_code and is more central than the narrow, model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "45020 is the best standalone anchor for this cluster because it is the most general regression report with the widest potential overlap to adjacent loading failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44496", + "right": "issue:44877", + "accept": false, + "reason": "Both are loading/config errors, but one is an unrecognized-model/config.json problem and the other is strict-config handling for granite_speech; different failure modes and fixes." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json is unrelated to the Mistral regex patch crash; different code paths and symptoms." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Video-classification processor lookup is not the same as broad remote_code model-loading regressions." + }, + { + "left": "issue:43550", + "right": "issue:44964", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba is distinct from Phi-4 multimodal loading issues." + }, + { + "left": "issue:44297", + "right": "issue:44462", + "accept": false, + "reason": "Tokenizer save metadata mismatch is a serialization bug; AutoTokenizer ignoring tokenizer.json is a load-path selection bug." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Generic dependency issue is unrelated to a PEFT minimum-version bump." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Two separate model-loading regressions for different models with no shared concrete code path evident." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Processor/tokenizer API regression is unrelated to Deepseek tokenizer output changes." + }, + { + "left": "issue:38175", + "right": "issue:44617", + "accept": false, + "reason": "SigLIP2 probability issue and Sam3Video OOM are unrelated model/runtime problems." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Docs/example errors for SigLIP2 are not the same as TextClassificationPipeline doc mismatch." + }, + { + "left": "issue:43650", + "right": "issue:45290", + "accept": false, + "reason": "Placeholder issue title is not related to chat-template tool-call crashes." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error and init_empty_weights argument regression are unrelated." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Both mention weights/embeddings, but one is serialization inconsistency and the other is resize_token_embeddings propagation; not the same bug." + }, + { + "left": "issue:43450", + "right": "issue:44279", + "accept": false, + "reason": "Batched video processor shape bug is unrelated to a dependency/version issue." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading regression and ESM2 model breakage are different issues." + }, + { + "left": "issue:43262", + "right": "issue:44521", + "accept": false, + "reason": "Both involve apply_chat_template, but one is default sampling rate and the other is assistant mask generation for multimodal inputs." + }, + { + "left": "issue:43441", + "right": "issue:44683", + "accept": false, + "reason": "FlashAttention failure for Ministral-3 and compiled flex_attention failure on torch>=2.9 are different attention-stack bugs." + }, + { + "left": "issue:43065", + "right": "issue:44898", + "accept": false, + "reason": "Sam3PixelDecoder dummy conv and Perceiver resize/interpolate failure are unrelated architecture-specific issues." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Quantized model saving failure and dequantized offload loading failure are related only at a high level, not the same underlying bug." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "MLA layernorm eps precision issue and GlmMoeDsaIndexer split behavior are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Missing processor.save_pretrained files and gemma-4 zero3 from_pretrained are separate save/load problems." + }, + { + "left": "issue:43122", + "right": "issue:43650", + "accept": false, + "reason": "Tokenizer behavior change across versions is unrelated to the placeholder 'ADD THE DATA' issue." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Processor save files missing and Chinese performance regression are unrelated." + }, + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Missing nn import in sentence-transformers and undefined tf backend when only tensorboard is installed are different optional-dependency failures." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Masked LM prediction regression and Deepseek tokenizer regression affect different components and code paths." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Generation cache missing in chunked_prefill and CLIPVision hidden_states regression are unrelated model behaviors." + }, + { + "left": "issue:44987", + "right": "issue:45542", + "accept": false, + "reason": "Physical-intelligence/fast loading failure is unrelated to the TensorFlow backend dependency 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": 13, + "review_unit_index": 29, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36246", + "issue:41628", + "issue:41950", + "issue:42175", + "issue:42491", + "issue:42757", + "issue:43295", + "issue:43299", + "issue:43352", + "issue:43450", + "issue:43475", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43673", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:43994", + "issue:44016", + "issue:44162", + "issue:44188", + "issue:44220", + "issue:44246", + "issue:44295", + "issue:44297", + "issue:44336", + "issue:44351", + "issue:44373", + "issue:44462", + "issue:44485", + "issue:44496", + "issue:44561", + "issue:44589", + "issue:44617", + "issue:44661", + "issue:44871", + "issue:45003", + "issue:45200", + "issue:45250", + "issue:45276", + "issue:45356", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561" + ], + "soft_pairs": [ + "issue:43901|issue:43994", + "issue:44297|issue:45397", + "issue:42175|issue:43976", + "issue:43475|issue:44016", + "issue:43824|issue:43976", + "issue:43352|issue:45250", + "issue:43295|issue:44617", + "issue:43525|issue:43881", + "issue:44589|issue:45362", + "issue:43450|issue:44617", + "issue:43901|issue:44246", + "issue:44188|issue:44336", + "issue:41628|issue:43976", + "issue:44295|issue:45397", + "issue:42491|issue:43299", + "issue:41950|issue:44220", + "issue:44336|issue:44373", + "issue:44246|issue:44496", + "issue:44462|issue:45356", + "issue:44661|issue:45542", + "issue:45003|issue:45561", + "issue:43994|issue:45276", + "issue:43526|issue:45003", + "issue:43531|issue:45362", + "issue:36246|issue:42757", + "issue:36246|issue:44351", + "issue:43673|issue:44561", + "issue:44336|issue:44485", + "issue:44162|issue:45200", + "issue:44246|issue:44871" + ], + "prepared_review_unit_hash": "2967174b8a6a71d9e66034282ad161cf8d29ca4a2799f5f9411e7614011313d7", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35861, + "estimated_input_tokens": 8966, + "estimated_eval_tokens": 18188 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: most pairs share a broad subsystem or release theme, but they do not look like the same underlying bug or change. I would not collapse any of the soft edges, and there is no clear canonical issue/PR for the whole set.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Both concern text-classification behavior, but one is a docs/behavior mismatch and the other is a model/pipeline correctness bug." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save metadata mismatch is unrelated to the Gemma-4 ZeRO-3 from_pretrained failure." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Different failures: missing TensorFlow backend in a torch install vs Python version compatibility breakage." + }, + { + "left": "issue:43475", + "right": "issue:44016", + "accept": false, + "reason": "SAM3 video attribute error and a notebook syntax error are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Import error for a missing model class is unrelated to Python 3.9/3.10 packaging support." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "One is a model-specific Flash Attention 2 incompatibility, the other is a generic Flash Attention topic." + }, + { + "left": "issue:43295", + "right": "issue:44617", + "accept": false, + "reason": "Processor/tokenizer regression is unrelated to the Sam3Video CUDA OOM report." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config and glm-4v-9b loading failure are different model-loading bugs." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to Qwen3.5-35B chat crashing." + }, + { + "left": "issue:43450", + "right": "issue:44617", + "accept": false, + "reason": "Incorrect batched video processor shape and CUDA OOM are distinct problems." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Docs mismatch in text-classification pipeline is not the same as intermittent slow import." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Torch.compile kernel divergence and ANSI logging output are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "Missing AutoImageProcessor import and Python-version incompatibility are different packaging issues." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "position_ids buffer read error is unrelated to Gemma-4 ZeRO-3 loading." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen3 MoE/VL, but the reported failures are different model/version compatibility issues." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video classification using image processors is unrelated to _torch_extract_fbank_features()." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "ANSI code leakage in loading_report and a wrong docstring are not the same bug." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Slow import and unrecognized model config are unrelated failures." + }, + { + "left": "issue:44462", + "right": "issue:45356", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json is not the same as the Kimi-K2.5 codec regression." + }, + { + "left": "issue:44661", + "right": "issue:45542", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES failure and undefined tf backend from tensorboard-only install are different dependency issues." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Unsafe sys.modules access and pytest-xdist file race are unrelated implementation bugs." + }, + { + "left": "issue:43994", + "right": "issue:45276", + "accept": false, + "reason": "Siglip2 nonsensical outputs and Gemma-4 embedding resize propagation are different model behaviors." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug and sys.modules access bug are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and Qwen3.5-35B chat crash are not the same issue." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Different missing-import regressions: Qwen2.5 VL image processor vs is_offline_mode from huggingface_hub." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they refer to different missing symbols and code paths." + }, + { + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Chunked-prefill cache issue and removal of is_torch_fx_available are separate v5 regressions." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "Loading-report ANSI leakage and GLM-5 RoPE implementation discussion are unrelated." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids defaulting are different model-specific bugs." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Intermittent import slowness and Gemma-3 eos_token_id inconsistency 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": 2, + "component_count": 13, + "review_unit_index": 30, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36246", + "issue:41628", + "issue:41950", + "issue:42886", + "issue:43065", + "issue:43066", + "issue:43232", + "issue:43257", + "issue:43262", + "issue:43295", + "issue:43408", + "issue:43441", + "issue:43475", + "issue:43540", + "issue:43596", + "issue:43606", + "issue:43653", + "issue:43673", + "issue:43901", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44336", + "issue:44368", + "issue:44466", + "issue:44496", + "issue:44734", + "issue:44749", + "issue:44779", + "issue:44857", + "issue:44877", + "issue:44933", + "issue:44977", + "issue:44987", + "issue:44995", + "issue:45003", + "issue:45137", + "issue:45200", + "issue:45229", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45459", + "issue:45563" + ], + "soft_pairs": [ + "issue:44242|issue:44857", + "issue:44779|issue:45292", + "issue:44295|issue:44749", + "issue:44877|issue:45375", + "issue:43441|issue:43540", + "issue:43066|issue:44779", + "issue:36246|issue:43596", + "issue:44987|issue:45459", + "issue:42886|issue:45229", + "issue:43901|issue:45003", + "issue:44995|issue:45200", + "issue:41628|issue:41950", + "issue:43257|issue:45137", + "issue:43065|issue:43408", + "issue:43295|issue:45292", + "issue:43232|issue:44336", + "issue:44496|issue:45003", + "issue:43262|issue:44246", + "issue:44060|issue:44077", + "issue:44466|issue:45335", + "issue:43475|issue:43540", + "issue:43653|issue:44162", + "issue:44368|issue:44977", + "issue:44368|issue:45563", + "issue:44933|issue:45459", + "issue:43441|issue:43606", + "issue:43673|issue:44734" + ], + "prepared_review_unit_hash": "7f116e26c142e9c7618f40136b634088718bc4f51c3856f74a0342a579bdcb5e", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35291, + "estimated_input_tokens": 8823, + "estimated_eval_tokens": 17902 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The proposed pairs are mostly only loosely related by subsystem or symptom. They do not look like the same concrete bug/change, so none are safe duplicate merges.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Different problems: MoE load-balancing loss vs an AMP/CUDA crash in image loss code." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Both touch tokenization/model embeddings, but one is a Deepseek tokenizer regression and the other is resize_token_embeddings not propagating to output embeddings." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "Unrelated symptoms: a position_ids buffer issue vs a performance regression after upgrading Transformers." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both mention config/strict handling, but they involve different models and different failure modes; not clearly the same bug." + }, + { + "left": "issue:43441", + "right": "issue:43540", + "accept": false, + "reason": "Different model paths and failures: FlashAttention on Ministral-3 vs video input processing in Qwen3OmniMoe." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer-related, but one is a decoder-type warning in v5 and the other is an incorrect Deepseek tokenizer result regression." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "ImportError for a specific Qwen image processor is unrelated to a DeepSpeed ZeRO3 BertModel index error." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Different concerns: model loading failure for a specific repo vs protobuf-related tokenizer error masking." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are unrelated issues." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Docs mismatch for a pipeline option vs unsafe sys.modules access in modeling_utils are unrelated." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "Different model-specific bugs: stale indexer cache on second forward pass vs mm_token_type_ids defaulting for Gemma 4 fine-tuning." + }, + { + "left": "issue:41628", + "right": "issue:41950", + "accept": false, + "reason": "ImportError for AutoImageProcessor and a video-classification pipeline looking for image processors are related only broadly, not the same concrete bug." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Both involve DeepSpeed/ZeRO3 contexts, but one is weight conversion with accelerate and the other is an empty-deque runtime error." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Both are SAM3-related, but one is a dummy Conv2d implementation and the other is a model-type mismatch warning." + }, + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "Both concern embedding/tokenizer interactions, but one is a regression around processor.tokenizer and images-to-tokenizer behavior, not resize_token_embeddings." + }, + { + "left": "issue:43232", + "right": "issue:44336", + "accept": false, + "reason": "Generation kwargs handling and ANSI codes in loading_report are unrelated." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model/config loading and unsafe sys.modules access are different failures." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults are unrelated to import latency." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "A tied-weights warning bug and a patchtsmixer post_init policy issue do not match." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Both involve weights/embeddings, but one is serialization depending on device and the other is resize_token_embeddings not updating decoder.embed_tokens." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "Different SAM3 video encoder attribute bug vs Qwen3OmniMoe video-processing ValueError." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are not the same issue." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "A tie_word_embeddings warning in Qwen3.5 fine-tuning is not the same as a flash-attention generation bug." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Warning about tie_word_embeddings is unrelated to a stale num_return_sequences warning in paged generate." + }, + { + "left": "issue:44933", + "right": "issue:45459", + "accept": false, + "reason": "Image_utils import problem and protobuf error masking are unrelated." + }, + { + "left": "issue:43441", + "right": "issue:43606", + "accept": false, + "reason": "Different code paths: FlashAttention failure vs device mismatch under CPU offload." + }, + { + "left": "issue:43673", + "right": "issue:44734", + "accept": false, + "reason": "Both involve generation/cache, but one is missing cache in chunked prefill and the other is wrong indexing in serve continuation; not clearly 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": 2, + "component_count": 13, + "review_unit_index": 31, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:41628", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42831", + "issue:42994", + "issue:43232", + "issue:43381", + "issue:43475", + "issue:43482", + "issue:43525", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43701", + "issue:43704", + "issue:43873", + "issue:43931", + "issue:44038", + "issue:44162", + "issue:44188", + "issue:44222", + "issue:44297", + "issue:44361", + "issue:44368", + "issue:44623", + "issue:44661", + "issue:44749", + "issue:44869", + "issue:44933", + "issue:44945", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45200", + "issue:45230", + "issue:45278", + "issue:45292", + "issue:45310", + "issue:45335", + "issue:45356", + "issue:45362", + "issue:45375", + "issue:45381", + "issue:45397", + "issue:45478" + ], + "soft_pairs": [ + "issue:44869|issue:45356", + "issue:43931|issue:44368", + "issue:43232|issue:43673", + "issue:44297|issue:44623", + "issue:43673|issue:45278", + "issue:43381|issue:45292", + "issue:43646|issue:44162", + "issue:43381|issue:45335", + "issue:44361|issue:44749", + "issue:28282|issue:30990", + "issue:42673|issue:43704", + "issue:44749|issue:45310", + "issue:45230|issue:45397", + "issue:44038|issue:44623", + "issue:43475|issue:43643", + "issue:42994|issue:43482", + "issue:43701|issue:44749", + "issue:43475|issue:45381", + "issue:42371|issue:45092", + "issue:45081|issue:45103", + "issue:45375|issue:45478", + "issue:41628|issue:44933", + "issue:45310|issue:45375", + "issue:42222|issue:45200", + "issue:43873|issue:44945", + "issue:42994|issue:43525", + "issue:43475|issue:44661", + "issue:42831|issue:44222", + "issue:45362|issue:45375", + "issue:45310|issue:45397", + "issue:43232|issue:44188", + "issue:44222|issue:45310" + ], + "prepared_review_unit_hash": "fd539651e5222e33ea58948b9935f39f51f9c32a6b6015f48e02a33864d308ba", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 36957, + "estimated_input_tokens": 9240, + "estimated_eval_tokens": 18736 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly unrelated issue noise, but it contains one clear duplicate pair about the same Qwen3ForCausalLM VRAM leak in multi-threaded dataloader use. No PRs are present.", + "confidence": 0.64, + "canonical_issue_reason": "issue:42673 is the strongest canonical issue: it is specific, concrete, and has an exact duplicate in issue:43704 with the same title and bug description.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:42673 is the best representative issue overall because it describes a clear, actionable bug and matches an exact duplicate; the rest of the cluster appears thematically mixed.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Different tokenizer/runtime failures: Whisper trailing replacement-character crash vs Kimi-K2.5 codec/fix_mistral_regex regression." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Different problems and models: Qwen3-VL weight-shape mismatch vs a Qwen3.5 warning about tie_word_embeddings during LoRA fine-tuning." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both are generation-related, but one is after sync_gpus and the other is missing cache during chunked prefill; not the same code-path bug." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both involve save_pretrained, but one is tokenizer metadata and the other is missing processor files; different artifacts and failures." + }, + { + "left": "issue:43673", + "right": "issue:45278", + "accept": false, + "reason": "A specific cache bug in chunked prefill is not the same as broad import errors after upgrading versions." + }, + { + "left": "issue:43381", + "right": "issue:45292", + "accept": false, + "reason": "Gradient checkpointing eval-mode restriction is unrelated to resize_token_embeddings not updating output embeddings." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Missing fields from trust_remote_code AutoConfig is unrelated to ESM2 being broken." + }, + { + "left": "issue:43381", + "right": "issue:45335", + "accept": false, + "reason": "Eval-mode gradient checkpointing issue is unrelated to t5gemma embedding resize not propagating." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "MLukeTokenizer AttributeError is unrelated to a Chinese-language performance regression." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "PyTorch import-missing error is not the same as Sentence Transformers getting stuck while loading." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Exact same title and bug: Qwen3ForCausalLM VRAM leak when used in multiple dataloader threads." + }, + { + "left": "issue:44749", + "right": "issue:45310", + "accept": false, + "reason": "Performance slowdown from upgrading is unrelated to Qwen3.5 Moe from_pretrained failure." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "Generic bug report vs a specific Gemma-4 zero3 from_pretrained problem; no concrete overlap." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe v5 bug is unrelated to processor.save_pretrained missing files." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "SAM 3 VisionEncoder missing attribute is unrelated to AutoConfig returning missing fields under trust_remote_code." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving failure is unrelated to Qwen2.5-GGUF loading failure." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "Checkpoint resume key mismatch is unrelated to the data filtering slowdown regression." + }, + { + "left": "issue:43475", + "right": "issue:45381", + "accept": false, + "reason": "Different multimodal bugs: SAM 3 missing attribute vs qwen2.5-vl video position IDs being wrong." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API guidance issue is unrelated to InternVL2 remote-code/meta-init incompatibility." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Both are AttributeErrors, but one is Mistral tokenizer regex patching and the other is auto_docstring annotations handling." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Related Qwen3.5 Moe family, but one is a missing config field in vision config and the other is a from_pretrained error; not enough to merge." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Missing AutoImageProcessor import and nonexistent image_utils import are different import-registry issues." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Both concern Qwen3.5 Moe, but one is a from_pretrained error and the other is a dropped vision config field; not the same bug." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "VitPose model breakage is unrelated to Gemma 4 mm_token_type_ids defaulting." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Quantization offloading issue is unrelated to incorrect outputs under pipeline parallelism." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized save failure is unrelated to Llama4Config missing pad_token_id." + }, + { + "left": "issue:43475", + "right": "issue:44661", + "accept": false, + "reason": "SAM 3 output attribute error is unrelated to add-new-model-like failing for tokenizer-mapped models." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression is not the same as FP8 save_pretrained failure for moe models." + }, + { + "left": "issue:45362", + "right": "issue:45375", + "accept": false, + "reason": "Qwen3.5-35B chat crash is unrelated to a missing deepstack_visual_indexes field in a different config class." + }, + { + "left": "issue:45310", + "right": "issue:45397", + "accept": false, + "reason": "Qwen3.5 Moe from_pretrained error is unrelated to Gemma-4 zero3 from_pretrained behavior." + }, + { + "left": "issue:43232", + "right": "issue:44188", + "accept": false, + "reason": "Both mention generation, but one is a sync_gpus kwargs update bug and the other is divergent attention kernels under torch.compile." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "FP8 save_pretrained for moe models is unrelated to Qwen3.5 Moe from_pretrained loading error." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: it correctly identifies a single exact duplicate pair (issue:42673 / issue:43704) and appropriately characterizes the rest as mostly unrelated. The canonical/best-issue rationale is conservative, and there are no PRs in the packet. The soft-edge rejections are also cautious and do not overclaim equivalence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 32, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:30990", + "issue:35141", + "issue:36010", + "issue:41628", + "issue:42175", + "issue:42898", + "issue:42994", + "issue:43232", + "issue:43278", + "issue:43377", + "issue:43479", + "issue:43504", + "issue:43526", + "issue:43673", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43874", + "issue:43976", + "issue:44038", + "issue:44060", + "issue:44077", + "issue:44079", + "issue:44164", + "issue:44188", + "issue:44242", + "issue:44351", + "issue:44373", + "issue:44448", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44792", + "issue:44869", + "issue:44933", + "issue:44964", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45137", + "issue:45276", + "issue:45310", + "issue:45356", + "issue:45459", + "issue:45561" + ], + "soft_pairs": [ + "issue:41628|issue:44351", + "issue:43504|issue:45561", + "issue:44496|issue:45561", + "issue:42994|issue:44164", + "issue:43526|issue:45561", + "issue:43232|issue:44792", + "issue:44038|issue:45310", + "issue:43377|issue:44242", + "issue:30990|issue:35141", + "issue:43673|issue:44188", + "issue:36010|issue:41628", + "issue:43479|issue:44079", + "issue:42898|issue:44568", + "issue:44964|issue:45137", + "issue:43479|issue:44869", + "issue:43976|issue:44933", + "issue:42175|issue:44933", + "issue:43784|issue:43824", + "issue:43278|issue:45137", + "issue:44623|issue:45310", + "issue:44188|issue:44373", + "issue:44060|issue:45125", + "issue:44734|issue:45103", + "issue:42898|issue:45356", + "issue:45137|issue:45276", + "issue:43723|issue:44568", + "issue:42898|issue:44779", + "issue:43874|issue:45092", + "issue:44077|issue:44242", + "issue:42898|issue:44448", + "issue:41628|issue:45459" + ], + "prepared_review_unit_hash": "7ddd3724fee4d8ab7da37e644fd8cab6b22a2bb9013b329f40723a7716aa0c6c", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36481, + "estimated_input_tokens": 9121, + "estimated_eval_tokens": 18498 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is highly heterogeneous: most items are distinct regression reports across unrelated model families, tokenizer paths, config/loading code, and test/infrastructure issues. No soft edge looks like a true duplicate pair.", + "confidence": 0.18, + "canonical_issue_reason": "42898 is the broadest, most representative issue in the set because it frames a v5 behavior regression with external reference activity, but the cluster overall does not form a cohesive duplicate group.", + "canonical_pr_reason": null, + "best_issue_reason": "42898 is the best available anchor for triage, though it still only represents one specific tokenizer-behavior regression and not the rest of the unrelated issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols and likely different API breakages." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "Semantic segmentation preset loading and a pytest-xdist file race are unrelated." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Unrecognized model config loading is unrelated to a testing-utils race condition." + }, + { + "left": "issue:42994", + "right": "issue:44164", + "accept": false, + "reason": "Both concern saving/loading, but one is quantized model saving and the other is extra_state serialization handling." + }, + { + "left": "issue:43526", + "right": "issue:45561", + "accept": false, + "reason": "A BEiT image processor label-reduction bug is unrelated to a captured_info.txt race." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation kwargs handling and a Janus image-generation test failure do not match the same underlying bug." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Both mention Qwen/Moe, but they appear to be different model/version-specific from_pretrained failures." + }, + { + "left": "issue:43377", + "right": "issue:44242", + "accept": false, + "reason": "Missing padding-mask support in MIMI encoder is unrelated to load-balancing loss wiring." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence Transformers loading hangs and embedding reinitialization after resize are different problems." + }, + { + "left": "issue:43673", + "right": "issue:44188", + "accept": false, + "reason": "GenerationMixin cache issues and attention-kernel divergence under torch.compile are separate code paths." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they concern different classes and different failing imports." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Config default initialization and ModelOutput key assignment are different object-model bugs." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but they affect different behaviors and different tokenizer code paths." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque underflow are unrelated." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Multimodal config defaults and Whisper timestamp decoding are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python-version compatibility and a missing image_utils import are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Packaging backend extras and a missing import symbol are unrelated." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "These are different import failures in different packages/symbols." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Embedding dtype drift during eval is unrelated to DeepSpeed ZeRO3 deque errors." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "Processor save-pretrained file omissions and Qwen3.5 MoE from_pretrained errors are different issues." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "Torch.compile attention-kernel divergence and a docstring issue are unrelated." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "A tied-weights warning bug and missing tensor-parallel plan metadata are different model-integration issues." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "KV-cache continuation indexing and auto-docstring annotation handling are unrelated." + }, + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they concern different models and different regression causes." + }, + { + "left": "issue:45137", + "right": "issue:45276", + "accept": false, + "reason": "DeepSpeed deque underflow and Gemma4 resize-token-embeddings behavior are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer loading failure and missing BOS/EOS insertion are different tokenizer issues." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 tokenizer regressions, but they affect different tokenizers and different behavior." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Missing multimodal helper method and remote-code/meta-init incompatibility are distinct multimodal problems." + }, + { + "left": "issue:44077", + "right": "issue:44242", + "accept": false, + "reason": "Optional post_init handling and load-balancing loss computation are not the same bug." + }, + { + "left": "issue:42898", + "right": "issue:44448", + "accept": false, + "reason": "Tokenizer cleanup behavior changes and Pegasus output drift are different regression reports." + }, + { + "left": "issue:41628", + "right": "issue:45459", + "accept": false, + "reason": "Missing top-level import and tokenizer protobuf-error masking are different import/tokenizer issues." + } + ] + }, + "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": 13, + "review_unit_index": 33, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:30990", + "issue:38175", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42898", + "issue:42994", + "issue:43066", + "issue:43122", + "issue:43295", + "issue:43479", + "issue:43576", + "issue:43644", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:44038", + "issue:44117", + "issue:44206", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44442", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44521", + "issue:44623", + "issue:44704", + "issue:44871", + "issue:44913", + "issue:44977", + "issue:44991", + "issue:45092", + "issue:45103", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45406", + "issue:45459" + ], + "soft_pairs": [ + "issue:44222|issue:44297", + "issue:42898|issue:43295", + "issue:44442|issue:45459", + "issue:43479|issue:43644", + "issue:44222|issue:44623", + "issue:42831|issue:44623", + "issue:44206|issue:45103", + "issue:22355|issue:28282", + "issue:44297|issue:45310", + "issue:41628|issue:42757", + "issue:44483|issue:45406", + "issue:44222|issue:44484", + "issue:44263|issue:44360", + "issue:43723|issue:44977", + "issue:44263|issue:45092", + "issue:44514|issue:45103", + "issue:44295|issue:44623", + "issue:43479|issue:44913", + "issue:44373|issue:44485", + "issue:43723|issue:44871", + "issue:44261|issue:45092", + "issue:43723|issue:44991", + "issue:43723|issue:44442", + "issue:30990|issue:38175", + "issue:45092|issue:45357", + "issue:42886|issue:42994", + "issue:42831|issue:45310", + "issue:41628|issue:43824", + "issue:42831|issue:44038", + "issue:43576|issue:43784", + "issue:43122|issue:44521", + "issue:44263|issue:44373", + "issue:42757|issue:45278", + "issue:44117|issue:44704", + "issue:42886|issue:43066", + "issue:43122|issue:44514" + ], + "prepared_review_unit_hash": "93be64134c587c9f16a8bcedef4e3239c2da9cdff8161620e7b5e5f702431282", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 38100, + "estimated_input_tokens": 9525, + "estimated_eval_tokens": 19306 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous, but a few issues sit around a shared v5 tokenizer-loading regression theme. Only one soft pair looks like a plausible duplicate; the rest are different bugs or different code paths.", + "confidence": 0.61, + "canonical_issue_reason": "issue:43723 is the broadest representative of the tokenizer-loading regressions in v5, and it can plausibly absorb specific model-level reports in that area.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43723 is the best single issue for the dominant subtheme (tokenizer loading failures), though it is not a true umbrella for the whole mixed cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "FP8/MoE save_pretrained failure vs tokenizer_class metadata mismatch; different bugs." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Tokenizer cleanup behavior change vs processor/tokenizer image-handling regression; different code paths." + }, + { + "left": "issue:44442", + "right": "issue:45459", + "accept": false, + "reason": "AutoTokenizer cannot load a specific tokenizer vs protobuf masking real tokenizer errors; not the same failure." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Default config initialization bug vs non-persistent buffer corruption; unrelated." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "FP8/MoE save_pretrained failure vs missing processor save files; different symptoms and causes." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "Accuracy regression in FP8 vs processor save_pretrained file omission; unrelated." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Auto-docstring crash with feature extractor args vs AttributeError under future annotations; same area, but different concrete bugs." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "ONNX module import failure vs missing PyTorch dependency error; unrelated." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer save metadata mismatch vs Qwen3.5 MoE from_pretrained failure; different issues." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Missing transformers export for AutoImageProcessor vs missing is_offline_mode import from huggingface_hub; different packages and failures." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Chat/completions request acceptance bug vs Gemma4Processor missing _tokenizer; different components." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "FP8 save_pretrained bug vs max_shard_size question; not duplicates." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "torch.split return handling vs missing ReLU in DSA indexer; different concrete bugs." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Generic tokenizer loading regression vs Qwen3.5 flash-attention generation bug; different paths." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "DSA indexer split bug vs InternVL2 meta-init incompatibility; unrelated." + }, + { + "left": "issue:44514", + "right": "issue:45103", + "accept": false, + "reason": "Qwen2_5_VLProcessor batched chat-template crash vs auto_docstring future-annotations crash; unrelated." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "position_ids buffer read error vs processor.save_pretrained missing files; different issues." + }, + { + "left": "issue:43479", + "right": "issue:44913", + "accept": false, + "reason": "Phi4MultimodalConfig defaulting bug vs GPTNeoXConfig rotary_pct reload bug; different configs." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Docstring text error vs GLM-5 RoPE implementation bug; not the same." + }, + { + "left": "issue:43723", + "right": "issue:44871", + "accept": false, + "reason": "Tokenizer loading regression vs eos_token_id config inconsistency; different model issues." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "Missing rms_norm_eps in MLA q_a_layernorm vs InternVL2 meta-init incompatibility; unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44991", + "accept": true, + "reason": "Both describe the same v5 AutoTokenizer loading regression; 44991 is a specific model instance of the broader loading failure." + }, + { + "left": "issue:43723", + "right": "issue:44442", + "accept": false, + "reason": "Generic tokenizer loading regression vs FastSpeech2ConformerTokenizer-specific load failure; related area but not clearly the same bug." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Sentence-Transformers loading hang vs SIGLIP2 zero-probability output; unrelated." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Meta initialization incompatibility vs Qwen3.5 save_pretrained visual key regression; different bugs." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Offline tokenizer cache load failure vs quantized model save failure; unrelated." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "FP8 accuracy issue vs Qwen3.5 MoE from_pretrained error; different failures." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Different missing imports for different classes; likely separate export issues." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "FP8 accuracy regression vs Qwen3-VL-Moe bug; unrelated." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "transformers env CLI regression vs sentence-transformers import NameError; different code paths." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "General tokenization output changes vs multimodal assistant_mask bug; different symptoms." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "Indexing bug vs docstring typo; unrelated." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "Specific missing import from huggingface_hub vs broad import-errors report; same broad area but not same concrete bug." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES None handling vs AutoProcessor kwargs propagation; different issues." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache tokenizer load failure vs wrong tokenizer decoder type in v5; different bugs." + }, + { + "left": "issue:43122", + "right": "issue:44514", + "accept": false, + "reason": "Version-to-version tokenization difference vs batched apply_chat_template crash; different failure modes." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The packet does show a heterogeneous cluster with a tokenizer-loading subtheme, and issue:43723 is a reasonable broad representative for that subtheme. The lone accepted soft pair (43723/44991) is plausible, while the other false verdicts stay safely on the conservative side rather than overclaiming duplicates." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 34, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:29127", + "issue:30990", + "issue:33357", + "issue:36296", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42757", + "issue:42831", + "issue:42907", + "issue:43116", + "issue:43122", + "issue:43232", + "issue:43278", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43824", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44360", + "issue:44373", + "issue:44485", + "issue:44488", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44743", + "issue:44779", + "issue:44829", + "issue:44945", + "issue:44964", + "issue:44977", + "issue:45200", + "issue:45216", + "issue:45245", + "issue:45278", + "issue:45290", + "issue:45478" + ], + "soft_pairs": [ + "issue:30990|issue:33357", + "issue:44704|issue:45290", + "issue:42371|issue:44263", + "issue:43232|issue:45245", + "issue:42907|issue:44945", + "issue:44295|issue:44297", + "issue:43232|issue:44743", + "issue:43976|issue:44945", + "issue:39401|issue:44488", + "issue:42907|issue:43937", + "issue:43408|issue:43475", + "issue:44625|issue:44829", + "issue:41628|issue:45278", + "issue:44360|issue:44485", + "issue:29127|issue:30990", + "issue:43116|issue:45200", + "issue:39401|issue:44521", + "issue:44261|issue:44373", + "issue:43122|issue:44779", + "issue:44038|issue:44297", + "issue:42371|issue:42831", + "issue:44038|issue:44295", + "issue:42907|issue:44964", + "issue:36296|issue:43116", + "issue:42907|issue:43452", + "issue:44038|issue:44222", + "issue:43992|issue:44704", + "issue:43122|issue:44977", + "issue:45216|issue:45478", + "issue:43278|issue:44945", + "issue:43824|issue:44351", + "issue:42907|issue:44222", + "issue:42757|issue:44351" + ], + "prepared_review_unit_hash": "dc54e8d59f280f0a43d590774fc4b3523f3a06f50bebe9ed567df94197419f71", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 36519, + "estimated_input_tokens": 9130, + "estimated_eval_tokens": 18516 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous. Most items are unrelated model-specific bugs or regressions that only share vague Transformer-version or import-related themes, so there are no strong duplicate pairs. The closest umbrella issue is the broad import-error report.", + "confidence": 0.9, + "canonical_issue_reason": "Issue 45278 is the broadest symptom-level report in the set, covering multiple import failures after an upgrade and making it the best umbrella for the small import-regression subgroup.", + "canonical_pr_reason": null, + "best_issue_reason": "45278 is the most triage-friendly general report here; it is broad enough to subsume several specific import errors, though it does not represent the many unrelated non-import bugs in the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Different failures: loading hang vs Mac bus error on a CLIP model. Same vague area, not the same bug." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "AutoProcessor kwarg forwarding and chat-template crashes are different code paths and different symptoms." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 configuration guidance and a GlmMoeDsaIndexer torch.split issue are unrelated." + }, + { + "left": "issue:43232", + "right": "issue:45245", + "accept": false, + "reason": "Generation sync/caching bug vs category-cardinality runtime error; no concrete overlap." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Saving dequantized models and incorrect pipeline-parallel outputs are different problems." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Reading a registered buffer and tokenizer_class metadata mismatch are separate serialization issues." + }, + { + "left": "issue:43232", + "right": "issue:44743", + "accept": false, + "reason": "Both touch generation/cache behavior, but they describe different failures in different paths." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version compatibility and pipeline-parallel output corruption are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Tokenizer offset mapping regression and model loading failure are not the same bug." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "FP8 accuracy concerns and a GenerationConfig validation error do not match." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Model-type warning and missing attribute in SAM 3 video are related only by subsystem, not by identical cause." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "num_labels propagation and flash_attention_3 training degeneration are different model/config issues." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "A single missing import and a broad upgrade-wide import-error report are not the same concrete defect." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "DSA indexer activation behavior and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:29127", + "right": "issue:30990", + "accept": false, + "reason": "Clarity of an error message vs a loading hang are not duplicate symptoms." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Empty multi-label outputs in an example script and Gemma 4 mm_token_type_ids defaults are different issues." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Offset mapping bug and all-zero assistant masks in multimodal chat templating are separate tokenizer/template problems." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "Missing RMS norm epsilon and a wrong docstring for position_ids are unrelated." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe loading in v5.0 and tokenizer save metadata mismatch are not the same bug." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "TF32 settings guidance and FineGrainedFP8 accuracy loss are different topics." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Model loading/regression and position_ids buffer access are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Quantized/dequantized save failure and Phi-4 multimodal loading failure do not share a concrete code-path." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training bug and multi-label example script output bug are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Saving dequantized models and gguf_file loading via AutoTokenizer/AutoModel are different serialization paths." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Qwen3-VL-Moe load bug and FP8 moe save_pretrained issue are different operations and likely different fixes." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Missing embed_tokens.weight in UMT5Encoder loading and AutoProcessor kwarg forwarding are unrelated." + }, + { + "left": "issue:43122", + "right": "issue:44977", + "accept": false, + "reason": "Different tokenizer regressions, on different models and with different symptoms." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Save_pretrained checkpoint corruption and a from_pretrained error are not enough to treat as the same concrete defect." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Dtype drift between train/eval and incorrect pipeline-parallel outputs are separate bugs." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Two import errors, but for different symbols and likely different causes." + }, + { + "left": "issue:42907", + "right": "issue:44222", + "accept": false, + "reason": "Both involve saving models, but one is dequantized Ministrals/Devstrals and the other is FP8 MoE; not the same bug." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Importing is_offline_mode from huggingface_hub and importing HybridCache from transformers are unrelated import failures." + } + ] + }, + "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": 13, + "review_unit_index": 35, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:22355", + "issue:29127", + "issue:33453", + "issue:38617", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:43064", + "issue:43116", + "issue:43421", + "issue:43441", + "issue:43450", + "issue:43493", + "issue:43550", + "issue:43673", + "issue:43784", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43906", + "issue:43931", + "issue:43937", + "issue:43976", + "issue:43986", + "issue:44038", + "issue:44075", + "issue:44222", + "issue:44291", + "issue:44484", + "issue:44485", + "issue:44779", + "issue:44829", + "issue:44936", + "issue:44945", + "issue:44977", + "issue:45071", + "issue:45072", + "issue:45083", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45405", + "issue:45446", + "issue:45478", + "issue:45561", + "issue:45563" + ], + "soft_pairs": [ + "issue:42222|issue:43441", + "issue:43784|issue:45446", + "issue:42673|issue:45072", + "issue:43064|issue:43673", + "issue:22355|issue:29127", + "issue:43550|issue:43784", + "issue:38617|issue:45278", + "issue:43937|issue:44485", + "issue:43116|issue:43493", + "issue:42371|issue:43874", + "issue:44075|issue:44936", + "issue:44222|issue:45478", + "issue:43116|issue:43825", + "issue:44038|issue:45478", + "issue:43976|issue:45446", + "issue:43937|issue:45563", + "issue:39401|issue:44779", + "issue:43931|issue:44945", + "issue:33453|issue:39401", + "issue:45083|issue:45446", + "issue:43937|issue:44977", + "issue:45341|issue:45561", + "issue:43827|issue:45446", + "issue:42673|issue:43906", + "issue:44291|issue:45071", + "issue:45405|issue:45446", + "issue:43450|issue:43986", + "issue:44829|issue:45216", + "issue:39401|issue:43421", + "issue:44484|issue:45310" + ], + "prepared_review_unit_hash": "0a3823f6403cfd829a1050a8c08a62ac37a73db301b8211b8d8abf28b269d4f2", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36397, + "estimated_input_tokens": 9100, + "estimated_eval_tokens": 18456 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated, model-specific regressions. The only near-duplicate I see is the configuration_utils import failure being a specific instance of the broader v5 import-regression report.", + "confidence": 0.67, + "canonical_issue_reason": "No single issue covers the whole set; if one must anchor the cluster, issue 45278 is the broadest umbrella for the import-regression subset.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the most representative issue here because it is broad enough to subsume the specific configuration_utils import failure in 38617.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Different models and failure modes: vitpose import breakage vs Ministral flash-attention error." + }, + { + "left": "issue:43784", + "right": "issue:45446", + "accept": false, + "reason": "Both are import-time problems, but they affect different modules and unrelated code paths." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3 CausalLM threads is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Training/FSDP optimizer-state corruption and chunked-prefill cache absence are different bugs." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "Module import failure for transformers.onnx is unrelated to a LayoutLMv3 error-message clarity issue." + }, + { + "left": "issue:43550", + "right": "issue:43784", + "accept": false, + "reason": "torch.compile+SDPA failure for Bamba is unrelated to the sentence-transformers import NameError." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": true, + "reason": "38617 is a specific import error that fits the broader v5 import-regression report in 45278." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "GLM-5 generation-config validation and RoPE implementation are different parts of the model stack." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Multi-label example-script output bug is unrelated to SigLIP2 implementation discrepancy." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 API guidance and missing image-patch helper are unrelated." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "SGD argument handling and trainer.evaluate() failure after train() are different code paths." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "save_pretrained FP8/moe behavior is not the same bug as from_pretrained loading failure." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Classification example output is unrelated to pipeline translation-task messaging after v5 removals." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen3-VL/Qwen3.5 MoE loading issues with no shared concrete failure described." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python version compatibility and PyTorch version gating are unrelated problems." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "GenerationConfig invalid and stale num_return_sequences warning are distinct generation issues." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Both involve tokenizers, but the titles point to different regressions and different models." + }, + { + "left": "issue:43931", + "right": "issue:44945", + "accept": false, + "reason": "Qwen3-VL weight-shape loading failure is unrelated to pipeline-parallel output corruption." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Tokenizer loading regression and Qwen3 offset_mapping bug are too different to treat as the same defect." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "Feature-extraction length helper behavior in qwen3_omni_moe is unrelated to flex_attention import version checks." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "GenerationConfig validation and flash-attention generation failure are separate issues." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "Generic testing_utils bug and xdist race on captured_info.txt are not clearly the same underlying problem." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs still referencing pipeline() and a flex_attention import check are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:43906", + "accept": false, + "reason": "The isolated reproduction target is unspecified here and does not match the VRAM-leak thread issue." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "init_empty_weights loading TypeError and PretrainedConfig type-checking breakage are different regressions." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "Unreleased PEFT minimum-version bump is unrelated to flex_attention import gating." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Batched video-processor shape bug and missing torchvision crash are different problems." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "flash_attention_3 training degeneration and save_pretrained checkpoint corruption are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:43421", + "accept": false, + "reason": "Offset-mapping regression and runtime special-token post-processor update are plausibly related, but not clearly the same bug from the available evidence." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "max_shard_size discussion and Qwen3.5 from_pretrained failure are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is conservative and grounded in the packet. The only clearly defensible near-duplicate is 38617 under the broader import-regression umbrella of 45278; the other soft-pair rejections are appropriately cautious based on the issue titles alone." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 36, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:38617", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42371", + "issue:42617", + "issue:42673", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43122", + "issue:43335", + "issue:43408", + "issue:43441", + "issue:43452", + "issue:43577", + "issue:43582", + "issue:43673", + "issue:43742", + "issue:43874", + "issue:43931", + "issue:43937", + "issue:43986", + "issue:44038", + "issue:44117", + "issue:44220", + "issue:44261", + "issue:44263", + "issue:44368", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44493", + "issue:44496", + "issue:44829", + "issue:44933", + "issue:44987", + "issue:45003", + "issue:45071", + "issue:45081", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45459", + "issue:45464", + "issue:45478", + "issue:45542" + ], + "soft_pairs": [ + "issue:38617|issue:41628", + "issue:44117|issue:45081", + "issue:44410|issue:45310", + "issue:42673|issue:43582", + "issue:43408|issue:43742", + "issue:42673|issue:43577", + "issue:42222|issue:45542", + "issue:42371|issue:44261", + "issue:43937|issue:44496", + "issue:38617|issue:44220", + "issue:44483|issue:45464", + "issue:42175|issue:42222", + "issue:42617|issue:45341", + "issue:44933|issue:45278", + "issue:44484|issue:45478", + "issue:43066|issue:43122", + "issue:43673|issue:45341", + "issue:43097|issue:44493", + "issue:44410|issue:45478", + "issue:42222|issue:45459", + "issue:43874|issue:44261", + "issue:43441|issue:44829", + "issue:43986|issue:44484", + "issue:43012|issue:43408", + "issue:44263|issue:44485", + "issue:41628|issue:45071", + "issue:45071|issue:45542", + "issue:45003|issue:45341", + "issue:43931|issue:44410", + "issue:44117|issue:44987", + "issue:43097|issue:43452", + "issue:43986|issue:45341", + "issue:43335|issue:45071", + "issue:43931|issue:44038", + "issue:43408|issue:44368", + "issue:43874|issue:44263" + ], + "prepared_review_unit_hash": "3f7497958e5b91fa3e3fce3e60c22a5c56032f61782cf02f864755c7aec54ebb", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 38311, + "estimated_input_tokens": 9578, + "estimated_eval_tokens": 19412 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft pairs look like clear same-bug duplicates under a strict triage standard; most are only loosely related by broad subsystem or version-regression themes. The strongest umbrella issue in the set is the broad import-regression report, but the evidence here is still too weak to merge any pair.", + "confidence": 0.34, + "canonical_issue_reason": "issue:45278 is the broadest representative of the visible cluster: it summarizes multiple import failures after a Transformers upgrade and is more likely to act as the umbrella report than the narrower, symbol-specific import issues.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best single issue to keep because it is broad, actionable, and likely covers the wider import-regression surface better than the more specific symptom reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and different entry points; no evidence they share the same underlying bug." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "These hit different code paths: tokenizer mapping fallback logic versus a Mistral regex patch crash. Similar upgrade-era failure mode, but not the same bug." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Both involve Qwen-family model loading, but the reported failures are different and point to different model/config paths." + }, + { + "left": "issue:42673", + "right": "issue:43582", + "accept": false, + "reason": "One is a VRAM leak in threaded dataloading; the other is an Apple Silicon TypeError in a warmup helper. No shared code-path evidence." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "A SAM3 model-type warning and a MobileLLM loading KeyError are unrelated symptoms." + }, + { + "left": "issue:42222", + "right": "issue:45542", + "accept": false, + "reason": "VitPose breakage and a TensorFlow backend detection/install issue are separate problems." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API guidance and an RMS norm epsilon precision bug are unrelated." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "Different model-loading failures with different causes: invalid GenerationConfig versus unrecognized model config." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "An import error and an audio feature extraction issue do not look like the same underlying bug." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both concern chat/completions behavior, but one is a general API rejection and the other is a streaming inference failure; not enough to treat as the same bug." + }, + { + "left": "issue:42175", + "right": "issue:42222", + "accept": false, + "reason": "TensorFlow installation/backend packaging and VitPose model breakage are not the same issue." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "A runtime script failure and a testing utility bug are unrelated." + }, + { + "left": "issue:44933", + "right": "issue:45278", + "accept": false, + "reason": "Both mention import problems, but one is a specific missing import from image_utils while the other is a broad post-upgrade import regression; insufficient evidence of one shared root cause." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "These are unrelated: max_shard_size behavior versus a Qwen3.5 MoE from_pretrained failure." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Both involve tokenization changes in v5, but the reported symptoms are different enough that the shared root cause is not established." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Generation cache handling and a testing-utils bug do not match." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Removed tie_embeddings support and unexpected position-id keys are different config/modeling regressions." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen-family failures with different model variants and symptoms; not the same concrete bug." + }, + { + "left": "issue:42222", + "right": "issue:45459", + "accept": false, + "reason": "VitPose model breakage and tokenizer error masking from protobuf handling are unrelated." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Different multimodal/model config issues with no clear shared code-path." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Both mention FlashAttention, but one is a Ministral loading failure and the other is degenerate training; not the same fix." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "Video AutoProcessor import/torchvision behavior is unrelated to max shard sizing." + }, + { + "left": "issue:43012", + "right": "issue:43408", + "accept": false, + "reason": "A precision warning during compile and a SAM3 model-type warning are unrelated." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "A torch.split return-value issue and a GLM RoPE implementation discussion are not the same bug." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "An import error and a PretrainedConfig type-checking regression are distinct issues." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "Type-checking and TensorFlow backend-install behavior are unrelated." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "Unsafe sys.modules access and a testing utility bug are not the same code-path." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Both involve model loading, but one is a weight-shape mismatch and the other is missing projections in a different model family." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Both are from_pretrained-related failures, but the tokenizer mapping bug and the physical-intelligence/fast loading failure do not clearly share a root cause." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Tie-embeddings removal and gguf_file loading breakage are different regression classes." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "Video processor dependency handling and a testing utility bug are unrelated." + }, + { + "left": "issue:43335", + "right": "issue:45071", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and config type-checking are separate problems." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Same broad Qwen3-VL area, but different symptoms and likely different causes." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Different GLM-related failures with no evidence of the same underlying defect." + } + ] + }, + "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": 13, + "review_unit_index": 37, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:38617", + "issue:42757", + "issue:42886", + "issue:42907", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43352", + "issue:43381", + "issue:43502", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43618", + "issue:43824", + "issue:43825", + "issue:43873", + "issue:43901", + "issue:43906", + "issue:43976", + "issue:43986", + "issue:43992", + "issue:44038", + "issue:44062", + "issue:44117", + "issue:44261", + "issue:44315", + "issue:44351", + "issue:44368", + "issue:44380", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44655", + "issue:44704", + "issue:44829", + "issue:45071", + "issue:45103", + "issue:45245", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45479" + ], + "soft_pairs": [ + "issue:44829|issue:45245", + "issue:43986|issue:45372", + "issue:43976|issue:45071", + "issue:43531|issue:45341", + "issue:43064|issue:44315", + "issue:38617|issue:42757", + "issue:44261|issue:44485", + "issue:44410|issue:45375", + "issue:44062|issue:44117", + "issue:44704|issue:45103", + "issue:44038|issue:44410", + "issue:38617|issue:44351", + "issue:43986|issue:44117", + "issue:43502|issue:43992", + "issue:43097|issue:43873", + "issue:43992|issue:44484", + "issue:43824|issue:43986", + "issue:43066|issue:43906", + "issue:43352|issue:44380", + "issue:43825|issue:44655", + "issue:38617|issue:43097", + "issue:43097|issue:43901", + "issue:43012|issue:43550", + "issue:43012|issue:44368", + "issue:42886|issue:43381", + "issue:43208|issue:43550", + "issue:42907|issue:44655", + "issue:43525|issue:44117", + "issue:44380|issue:45479", + "issue:43352|issue:43976", + "issue:43064|issue:44380", + "issue:43208|issue:43526", + "issue:44483|issue:44514", + "issue:43618|issue:44380" + ], + "prepared_review_unit_hash": "0edf23d0cdcaf93b23a3e0bd2e5f665db51e3005c5b1359a0887f6b371aa6907", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37827, + "estimated_input_tokens": 9457, + "estimated_eval_tokens": 19170 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the soft pairs mostly connect unrelated regressions, import errors, model-specific bugs, and docs/API mismatches. I reject all soft edges and do not see a true duplicate cluster here.", + "confidence": 0.95, + "canonical_issue_reason": "No single issue is a clear duplicate anchor for the whole set. If forced to pick a representative, issue 43097 is the clearest user-facing v5 regression with broad downstream impact and a concrete symptom.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43097 is the strongest standalone representative: it is specific, reproducible, and affects a core v5 API change. It is the best candidate only as an anchor, not because the surrounding issues are duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44829", + "right": "issue:45245", + "accept": false, + "reason": "Different failures: flash-attention training degeneration vs a categorical cardinality limit error." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Both are loading/import problems, but they involve different missing dependencies and different processor paths." + }, + { + "left": "issue:43976", + "right": "issue:45071", + "accept": false, + "reason": "One is a Python version compatibility complaint; the other is a PretrainedConfig type-checking regression." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior is unrelated to a testing_utils bug." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "Trainer/FSDP optimizer-state corruption is a different code path from Liger Kernel not being applied with model_init." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Different import errors from different symbols/modules; same symptom class, not the same bug." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "Different model internals: MLA RMS norm precision issue vs GLM-5 RoPE implementation." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Both are Qwen3-family issues, but one is missing projections and the other is a missing config field." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "Tokenizer construction error vs tokenizer mapping fallback bug; not the same concrete failure." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwarg forwarding and auto_docstring annotation handling are unrelated code paths." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Distinct Qwen model bugs with different symptoms and fixes." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they refer to different missing public symbols." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Video-model processor loading without torchvision is unrelated to tokenizer mapping returning None." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "Local-files-only network leakage is unrelated to a missing weight during UMT5Encoder loading." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "Tie-embeddings deprecation/regression is unrelated to quantization offloading behavior." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing embed_tokens weight on load and save_pretrained shard-size behavior are different operations." + }, + { + "left": "issue:43824", + "right": "issue:43986", + "accept": false, + "reason": "Importing a missing model class is not the same issue as crashing on AutoProcessor without torchvision." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Tokenizer decoder-type regression is too different from an isolated reproduction of another issue." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Unsupported FlashAttention 2 warning for Nemotron is different from GPT2 attention scaling being ignored." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline translation-message wording and saving Pipeline objects are separate behaviors." + }, + { + "left": "issue:38617", + "right": "issue:43097", + "accept": false, + "reason": "ImportError for a missing helper and a removed config field are unrelated." + }, + { + "left": "issue:43097", + "right": "issue:43901", + "accept": false, + "reason": "Removed tie_embeddings_and_encoder_decoder and return_all_scores doc drift do not share the same bug." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "Compile-time warning on bfloat16 differs from a functional torch.compile+SDPA failure." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Precision warning during compile is unrelated to the tie_word_embeddings warning in LORA fine-tuning." + }, + { + "left": "issue:42886", + "right": "issue:43381", + "accept": false, + "reason": "Offline cache loading and eval-mode gradient checkpointing are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Saving dequantized models is a different problem from saving Pipeline objects." + }, + { + "left": "issue:43525", + "right": "issue:44117", + "accept": false, + "reason": "Llama4Config missing pad_token_id is unrelated to tokenizer mapping assumptions." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention backend scaling bug and degenerate classification loss with num_labels=1 are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Model backend support and Python version compatibility are not the same underlying issue." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "Optimizer-state corruption under FSDP2/PEFT is unrelated to GPT2 attention scaling." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training blockers and BeitImageProcessorFast label reduction are unrelated." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Server endpoint request rejection and processor chat-template batching crash are different layers and failures." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "CLIPOutput attention assignment regression is unrelated to GPT2 attention scaling under backend kernels." + } + ] + }, + "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": 13, + "review_unit_index": 38, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:36010", + "issue:36683", + "issue:38617", + "issue:41669", + "issue:42757", + "issue:42994", + "issue:43010", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43502", + "issue:43525", + "issue:43582", + "issue:43618", + "issue:43756", + "issue:43761", + "issue:43824", + "issue:43867", + "issue:43901", + "issue:43931", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44077", + "issue:44188", + "issue:44230", + "issue:44242", + "issue:44246", + "issue:44263", + "issue:44380", + "issue:44483", + "issue:44485", + "issue:44492", + "issue:44655", + "issue:44683", + "issue:44938", + "issue:45216", + "issue:45278", + "issue:45290", + "issue:45310", + "issue:45375", + "issue:45381", + "issue:45440", + "issue:45478" + ], + "soft_pairs": [ + "issue:43618|issue:43761", + "issue:36683|issue:43582", + "issue:44483|issue:45290", + "issue:43012|issue:44683", + "issue:44038|issue:44230", + "issue:42757|issue:43824", + "issue:44483|issue:45381", + "issue:43299|issue:43867", + "issue:38617|issue:43502", + "issue:43756|issue:44485", + "issue:43901|issue:44655", + "issue:42994|issue:44655", + "issue:36683|issue:43475", + "issue:43867|issue:45216", + "issue:43066|issue:45381", + "issue:43352|issue:43937", + "issue:44655|issue:45478", + "issue:44077|issue:44380", + "issue:36683|issue:44077", + "issue:44263|issue:45440", + "issue:43064|issue:44492", + "issue:43976|issue:44938", + "issue:36683|issue:43525", + "issue:36010|issue:42757", + "issue:44188|issue:44380", + "issue:43618|issue:45216", + "issue:36683|issue:44242", + "issue:43992|issue:45310", + "issue:43010|issue:45278", + "issue:36683|issue:43352", + "issue:43352|issue:43931", + "issue:41669|issue:44246", + "issue:43867|issue:45375", + "issue:36683|issue:43064" + ], + "prepared_review_unit_hash": "ceebbf64f0b69fdd80c7df3b57734f75e17a7b50f8308f14e9658971b7b7db0a", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37419, + "estimated_input_tokens": 9355, + "estimated_eval_tokens": 18966 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous issue cluster with one clear duplicate family around import-time slowdown. One soft pair looks like the same import-performance bug; the rest are separate model/config/version-specific regressions.", + "confidence": 0.71, + "canonical_issue_reason": "issue:45216 is the clearest concrete regression in the Qwen3.5 MoE save/load family and reads like the best anchor for nearby reports about the same checkpoint/from_pretrained failure mode.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45216 is the most specific, actionable report among the issues here; it names the affected model, API, and version boundary.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both are CLIP/v5 regressions, but one is about attentions and the other hidden_states; different concrete failures." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Unrelated failures: missing Gemma3Config vocab_size vs Apple Silicon caching allocator TypeError." + }, + { + "left": "issue:44483", + "right": "issue:45290", + "accept": false, + "reason": "Different paths: chat/completions request validation vs apply_chat_template crashing on tool-call messages." + }, + { + "left": "issue:43012", + "right": "issue:44683", + "accept": false, + "reason": "Both touch compilation/warnings, but one is a bfloat16 precision warning and the other is a torch>=2.9 flex_attention failure." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Same model family, but one is a loading bug and the other is an fp8 inference support request; not the same bug." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and different breakpoints." + }, + { + "left": "issue:44483", + "right": "issue:45381", + "accept": false, + "reason": "Different bugs: API request rejection vs Qwen2.5-VL video position ids being wrong." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Both are model-loading errors, but the affected models and failure modes differ." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "ImportError from configuration_utils is unrelated to local_files_only still allowing network requests." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Both mention RoPE, but they target different models and different implementation concerns." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mismatch for return_all_scores is unrelated to saving Pipeline objects." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Quantized model saving and Pipeline.save_pretrained are different save-path bugs." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Different attribute-missing regressions in different model configs/outputs." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Both are Qwen-related load/save problems, but one is a shape-mismatch load error and the other is a checkpoint correctness regression." + }, + { + "left": "issue:43066", + "right": "issue:45381", + "accept": false, + "reason": "Tokenizer decoder type issue is unrelated to Qwen2.5-VL video position ids." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Different model/runtime issues: Flash Attention support vs GenerationConfig validation." + }, + { + "left": "issue:44655", + "right": "issue:45478", + "accept": false, + "reason": "Pipeline save_pretrained and Qwen3.5 MoE from_pretrained are separate failure paths." + }, + { + "left": "issue:44077", + "right": "issue:44380", + "accept": false, + "reason": "Different code paths: post_init validation vs attention scaling under SDPA/FlashAttention." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Gemma3Config missing vocab_size is unrelated to patchtsmixer post_init rules." + }, + { + "left": "issue:44263", + "right": "issue:45440", + "accept": false, + "reason": "Different model implementations and different correctness issues." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "Distributed optimizer-state bug is unrelated to a cache-strategy typo." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both involve Python compatibility, but they are distinct version/load failures." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Different configs missing different attributes; not the same bug." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Separate import errors for different symbols and causes." + }, + { + "left": "issue:44188", + "right": "issue:44380", + "accept": false, + "reason": "Attention-kernel divergence under compile is a different bug from GPT2 attention scaling being ignored." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIP output regression is unrelated to Qwen3.5 save/load behavior." + }, + { + "left": "issue:36683", + "right": "issue:44242", + "accept": false, + "reason": "Missing config attribute vs MoE load-balancing loss omission are unrelated." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Different from_pretrained problems on different model families." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "Cache update no_grad decoration is unrelated to broad import errors after upgrade." + }, + { + "left": "issue:36683", + "right": "issue:43352", + "accept": false, + "reason": "Config attribute error and Flash Attention support error are different failures." + }, + { + "left": "issue:43352", + "right": "issue:43931", + "accept": false, + "reason": "Nemotron Flash Attention support is unrelated to Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": true, + "reason": "Same underlying import-time slowdown: one report identifies the model import-star cause, the other reports the symptom that importing transformers is slow." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Weight-shape mismatch on load is unrelated to a missing deepstack_visual_indexes field being dropped." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Different subsystems and different failure modes." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The only accepted soft pair (issue:41669 / issue:44246) is supported by the shared import-performance symptom, and the other pairwise rejections do not overclaim identity. The Qwen3.5 MoE regression anchor at issue:45216 is a reasonable choice for the nearby save/pretrained reports." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 39, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:31795", + "issue:33453", + "issue:36296", + "issue:42548", + "issue:42890", + "issue:43010", + "issue:43065", + "issue:43232", + "issue:43299", + "issue:43502", + "issue:43575", + "issue:43618", + "issue:43723", + "issue:43867", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44075", + "issue:44077", + "issue:44112", + "issue:44222", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44368", + "issue:44373", + "issue:44387", + "issue:44484", + "issue:44485", + "issue:44556", + "issue:44749", + "issue:44857", + "issue:44871", + "issue:44938", + "issue:44964", + "issue:44977", + "issue:45071", + "issue:45125", + "issue:45341", + "issue:45357", + "issue:45375", + "issue:45381", + "issue:45440", + "issue:45478", + "issue:45561" + ], + "soft_pairs": [ + "issue:42890|issue:45561", + "issue:43992|issue:45478", + "issue:44938|issue:45071", + "issue:43867|issue:45357", + "issue:43010|issue:43976", + "issue:44938|issue:45381", + "issue:44360|issue:45440", + "issue:43618|issue:44077", + "issue:36296|issue:43575", + "issue:43867|issue:44484", + "issue:43618|issue:44857", + "issue:28282|issue:33453", + "issue:43618|issue:44977", + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:43010|issue:44075", + "issue:43723|issue:44556", + "issue:42890|issue:45341", + "issue:43502|issue:44062", + "issue:44387|issue:44556", + "issue:44222|issue:45125", + "issue:43010|issue:43065", + "issue:43010|issue:43232", + "issue:45125|issue:45478", + "issue:42548|issue:44351", + "issue:45125|issue:45375", + "issue:44355|issue:44749", + "issue:43502|issue:44485", + "issue:30990|issue:31795", + "issue:44075|issue:44871", + "issue:44315|issue:44360", + "issue:43010|issue:44368", + "issue:42890|issue:44373", + "issue:43502|issue:44222", + "issue:43299|issue:44556" + ], + "prepared_review_unit_hash": "41410ec29671a3804d8b3b57045d3d98c5383941cac953ba796e4c44da143b35", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37642, + "estimated_input_tokens": 9411, + "estimated_eval_tokens": 19078 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a collection of superficially similar Transformers issues (loading, version regressions, tensor-parallel, docs, tests) but not duplicates of the same underlying bug. I would not merge any of the soft-edge pairs.", + "confidence": 0.95, + "canonical_issue_reason": "issue:43723 is the most broadly framed, concrete loading-regression report in the set (AutoTokenizer.from_pretrained in v5), making it the best representative issue among these superficially related reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43723 is a clear, actionable model-loading regression with a specific API surface and version context; it is the strongest general representative of the loading-related subset in this cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Different test bugs: one is about missing set_seed in SAM HQ integration tests, the other is a pytest-xdist race on captured_info.txt." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Both are loading-related, but one is a missing embed_tokens.weight in UMT5Encoder and the other is a Qwen3.5 MoE from_pretrained error; different model/code paths." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Unrelated issues: Python 3.14 load failure vs PretrainedConfig type-checking breakage in v5.4.0." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Different failure modes: sorted state_dict loading error vs incorrect visual encoder keys saved by Qwen3.5 save_pretrained." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "No shared bug: one is about @torch.no_grad on cache/layer updates, the other is a Python-version compatibility problem." + }, + { + "left": "issue:44938", + "right": "issue:45381", + "accept": false, + "reason": "Different model/version bugs: Python 3.14 import/load failure vs wrong vision_position_ids for qwen2.5-vl video input." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "Completely different areas: DSA indexer ReLU discussion vs DeepseekV3MoE implementation divergence." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Different components: CLIPOutput attentions regression vs patchtsmixer post_init behavior." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Both mention tensor parallelism, but one is a generic training bug and the other is an OOM while loading a specific Qwen2 model; not the same bug." + }, + { + "left": "issue:43867", + "right": "issue:44484", + "accept": false, + "reason": "Loading a sorted state_dict and save_pretrained max_shard_size semantics are unrelated." + }, + { + "left": "issue:43618", + "right": "issue:44857", + "accept": false, + "reason": "CLIP attentions regression and float16 AMP crash in LwDetrImageLoss are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "Different subsystems: missing PyTorch dependency for AutoModel import vs tokenizer loading regression." + }, + { + "left": "issue:43618", + "right": "issue:44977", + "accept": false, + "reason": "CLIPOutput attentions issue is unrelated to Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Missing set_seed in tests is unrelated to loading microsoft/Phi-4-multimodal-instruct." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/test-related, but one is set_seed coverage and the other is a stale device-override test failure; not the same bug." + }, + { + "left": "issue:43010", + "right": "issue:44075", + "accept": false, + "reason": "Different code paths: cache update no_grad decoration vs optimizer SGD args being ignored." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Both concern loading, but one is tokenizer loading in v5 and the other is reloading checkpoints after upgrade; too broad and not the same concrete defect." + }, + { + "left": "issue:42890", + "right": "issue:45341", + "accept": false, + "reason": "Unrelated test issues: set_seed omissions vs a bug in testing_utils.py." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "Different problems: local_files_only still triggering API requests vs tokenizers.AddedToken() receiving duplicate 'special' kwargs." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "OOM under int4 quantization is unrelated to checkpoint reload failures after upgrading from v4.57 to v5.x." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "Different MoE/tensor-parallel topics: FP8 save_pretrained versus missing _tp_plan for tensor parallelism." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "No overlap: no_grad decoration for cache updates vs a dummy Conv2d in Sam3PixelDecoder." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "Different generation/cache bugs; the titles point to separate code paths and symptoms." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "The Qwen3.5 MoE from_pretrained error may be related to tensor parallelism, but there is no evidence it is the same concrete bug as the missing _tp_plan issue." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import-name errors, but they involve different exported symbols (PreTrainedModel vs HybridCache) and likely different causes." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Both mention Qwen3.5 MoE-related config/loading concerns, but one is missing _tp_plan and the other is a dropped vision config field; different defects." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors and a slowdown after upgrading Transformers are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "local_files_only API leakage and max_shard_size defaults are unrelated save/load behaviors." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Sentence Transformers loading hang and a docstring clarification for forward args are not the same bug." + }, + { + "left": "issue:44075", + "right": "issue:44871", + "accept": false, + "reason": "Optimizer argument handling and Gemma eos_token_id config mismatch are unrelated." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Different topics entirely: Liger Kernel application in model_init vs a DSA indexer missing ReLU." + }, + { + "left": "issue:43010", + "right": "issue:44368", + "accept": false, + "reason": "Cache update no_grad and a warning about tie_word_embeddings=False are unrelated." + }, + { + "left": "issue:42890", + "right": "issue:44373", + "accept": false, + "reason": "Test seed flakiness and a wrong docstring for position_ids are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "API requests despite local_files_only and FP8 save_pretrained for MoE are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Qwen3VL MoE loading break in dev0 and generic checkpoint reload failures after upgrading are similar only at a high level, not the same concrete 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": 13, + "review_unit_index": 40, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:31795", + "issue:33357", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:38617", + "issue:41628", + "issue:41669", + "issue:42371", + "issue:42548", + "issue:42757", + "issue:42831", + "issue:42890", + "issue:43012", + "issue:43502", + "issue:43526", + "issue:43576", + "issue:43673", + "issue:43824", + "issue:43827", + "issue:43874", + "issue:43901", + "issue:44038", + "issue:44222", + "issue:44230", + "issue:44246", + "issue:44261", + "issue:44273", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44393", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44655", + "issue:44704", + "issue:44908", + "issue:44961", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45230", + "issue:45468", + "issue:45542" + ], + "soft_pairs": [ + "issue:43502|issue:44961", + "issue:44222|issue:44230", + "issue:44355|issue:44655", + "issue:44355|issue:45071", + "issue:42831|issue:43874", + "issue:33453|issue:34689", + "issue:42757|issue:43576", + "issue:43827|issue:44961", + "issue:33453|issue:36296", + "issue:43576|issue:45071", + "issue:43576|issue:43824", + "issue:43673|issue:44492", + "issue:38617|issue:43576", + "issue:33357|issue:33453", + "issue:44961|issue:45468", + "issue:31795|issue:38617", + "issue:43576|issue:44351", + "issue:44908|issue:44961", + "issue:41628|issue:42548", + "issue:36296|issue:43901", + "issue:44485|issue:44961", + "issue:36296|issue:43526", + "issue:36296|issue:43012", + "issue:41628|issue:44355", + "issue:44273|issue:45003", + "issue:44261|issue:44360", + "issue:44273|issue:45542", + "issue:44623|issue:45230", + "issue:43576|issue:44556", + "issue:41669|issue:43673", + "issue:44393|issue:45071", + "issue:44038|issue:45125", + "issue:44556|issue:45092", + "issue:42371|issue:44492", + "issue:42890|issue:44908", + "issue:44704|issue:44908", + "issue:44246|issue:44273" + ], + "prepared_review_unit_hash": "aae42b06cd365e0f357b9a57d63b6fdee80eb4e2200ddc1a6a471be399d6fe7f", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 37, + "serialized_chars": 37410, + "estimated_input_tokens": 9353, + "estimated_eval_tokens": 18962 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The items are largely unrelated regression/bug reports across different subsystems. No soft pair looks like the same underlying bug or change, so all proposed edges are rejected. The strongest standalone issue to anchor triage is the broad import-performance report.", + "confidence": 0.84, + "canonical_issue_reason": "If one issue must represent this set, issue 44246 is the best anchor because it is broad, user-visible, and actionable (import performance regression), unlike the more narrowly scoped or highly specialized reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44246 is the most generally applicable and triage-friendly: it affects core import behavior, has a clear symptom, and is not tied to a single niche model or feature path.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "Unrelated: local_files_only network leakage vs a placeholder/title-only issue." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "Both are FP8/MoE-related, but save_pretrained persistence and inference support are different code paths." + }, + { + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Different failures: running compiled Python files vs saving Pipeline objects." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Different bugs: compiled-file execution errors vs PretrainedConfig type-checking regression." + }, + { + "left": "issue:42831", + "right": "issue:43874", + "accept": false, + "reason": "Different subsystems: FP8 accuracy vs missing image-processor method." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Both are loading regressions, but they involve different models and different failure modes." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "Different issues: missing huggingface_hub import vs broken transformers env command." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs referencing pipeline() is unrelated to the placeholder issue." + }, + { + "left": "issue:33453", + "right": "issue:36296", + "accept": false, + "reason": "Tokenizer loading regression and tensor-parallel training bug are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "Both are v5 regressions, but command behavior and type checking are separate problems." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "Broken env command and missing Qwen2.5-VL export are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Cache-missing runtime bug vs a typo in cache strategy text; not the same issue." + }, + { + "left": "issue:38617", + "right": "issue:43576", + "accept": false, + "reason": "ImportError for layer_type_validation is unrelated to the env command regression." + }, + { + "left": "issue:33357", + "right": "issue:33453", + "accept": false, + "reason": "Different regressions affecting different loading paths and models." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "Placeholder issue is unrelated to Gemma-4 audio positional encoding." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation confusion vs an ImportError are not the same underlying bug." + }, + { + "left": "issue:43576", + "right": "issue:44351", + "accept": false, + "reason": "Broken env command and missing HybridCache import are separate breakages." + }, + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Scheduler kwargs bug is unrelated to the placeholder issue." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different re-export paths." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug vs pipeline docs mismatch are unrelated." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion is unrelated to the placeholder issue." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Tensor-parallel training and BeitImageProcessorFast label reduction are different bugs." + }, + { + "left": "issue:36296", + "right": "issue:43012", + "accept": false, + "reason": "Training bug vs bfloat16 compile warning are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:44355", + "accept": false, + "reason": "Different import/export failure and compiled-file execution problem." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Lazy loading symptoms may sound related, but the titles describe different failure modes and code paths." + }, + { + "left": "issue:44261", + "right": "issue:44360", + "accept": false, + "reason": "Precision error in MLA q_a_layernorm is unrelated to the DSA indexer/ReLU discussion." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading issue and TensorFlow backend error are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "Missing processor.save_pretrained files is not the same as a generic bug report." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Both are v5-related breakages, but a broken CLI/env command is not the same as checkpoint reload incompatibility." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-star cleanup/performance issue is unrelated to GenerationMixin cache behavior." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "Qwen3-VL bbox hallucination/error and PretrainedConfig type checking are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "General Qwen3-VL-Moe bug and missing _tp_plan for tensor parallelism are different concrete issues." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Both involve v5 compatibility, but they concern different checkpoint/model incompatibility causes." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 behavior settings and a cache-strategies typo are unrelated." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Test flakiness from missing set_seed() is unrelated to scheduler kwargs handling." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor cached_file kwargs passthrough and inverse_sqrt scheduler kwargs are different bugs." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import performance regression and lazy-loading malfunction are related in theme but not clearly 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": 2, + "component_count": 13, + "review_unit_index": 41, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:34689", + "issue:37428", + "issue:41669", + "issue:42371", + "issue:42913", + "issue:43066", + "issue:43352", + "issue:43408", + "issue:43502", + "issue:43519", + "issue:43525", + "issue:43576", + "issue:43606", + "issue:43618", + "issue:43704", + "issue:43723", + "issue:43827", + "issue:43976", + "issue:44077", + "issue:44261", + "issue:44263", + "issue:44273", + "issue:44295", + "issue:44297", + "issue:44315", + "issue:44351", + "issue:44360", + "issue:44393", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44655", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:44945", + "issue:44998", + "issue:45071", + "issue:45092", + "issue:45230", + "issue:45290", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45478" + ], + "soft_pairs": [ + "issue:43408|issue:44315", + "issue:44297|issue:45230", + "issue:43576|issue:44908", + "issue:43352|issue:44393", + "issue:43976|issue:44393", + "issue:44908|issue:45071", + "issue:44295|issue:45230", + "issue:41669|issue:44351", + "issue:41669|issue:45341", + "issue:44492|issue:44945", + "issue:41669|issue:44492", + "issue:44315|issue:44829", + "issue:44393|issue:45478", + "issue:44077|issue:44861", + "issue:43827|issue:44908", + "issue:31795|issue:33453", + "issue:41669|issue:44908", + "issue:43723|issue:44393", + "issue:44655|issue:44861", + "issue:43576|issue:44273", + "issue:44360|issue:45468", + "issue:45230|issue:45310", + "issue:44273|issue:44908", + "issue:41669|issue:43704", + "issue:43519|issue:44485", + "issue:43519|issue:44908", + "issue:31795|issue:34689", + "issue:31795|issue:37428", + "issue:44315|issue:45092", + "issue:43704|issue:44485", + "issue:44485|issue:45468", + "issue:42371|issue:43704", + "issue:43502|issue:43519", + "issue:43519|issue:44623", + "issue:43519|issue:43704", + "issue:44263|issue:45468", + "issue:44261|issue:45468", + "issue:43606|issue:45468", + "issue:43704|issue:45341", + "issue:42913|issue:43066", + "issue:44908|issue:44998", + "issue:44861|issue:45290", + "issue:44998|issue:45341", + "issue:44492|issue:44998", + "issue:43618|issue:44998", + "issue:44556|issue:44998", + "issue:43525|issue:44861" + ], + "prepared_review_unit_hash": "11e42a2b65ae3e14f8046cd1449bbbc79b34adc4ddafdbb3dee411ee01843896", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 47, + "serialized_chars": 41102, + "estimated_input_tokens": 10276, + "estimated_eval_tokens": 20808 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are clearly unrelated; the only plausible duplicate is the v5 tokenizer behavior pair (42913/43066). I would keep the specific decoder-type regression as the main issue target.", + "confidence": 0.81, + "canonical_issue_reason": "issue:43066 is the most concrete tokenizer/v5 regression here: it names the specific wrong decoder-type behavior, making it the clearest duplicate target over the broader symptom report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43066 is the strongest standalone issue in this set because it is specific, reproducible, and actionable; the broader tokenizer-behavior report fits beneath it.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "Different problems: model-type mismatch warning vs Liger Kernel application in model_init." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer save_pretrained bug vs a generic bug report; no clear same underlying failure." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "Unrelated: broken env command vs inverse_sqrt scheduler kwargs handling." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention 2.0 support error vs Qwen3-VL bbox hallucination; different code paths." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python compatibility regression vs Qwen3-VL bounding-box output bug." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwargs ignored vs PretrainedConfig type-checking breakage." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "Position_ids buffer access issue vs generic bug report; insufficient evidence they are the same." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-star cleanup/perf issue is unrelated to HybridCache import failure." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Different targets: model import cleanup vs testing_utils bug." + }, + { + "left": "issue:44492", + "right": "issue:44945", + "accept": false, + "reason": "Cache-strategy typo is unrelated to pipeline-parallelism output correctness." + }, + { + "left": "issue:41669", + "right": "issue:44492", + "accept": false, + "reason": "Broad import cleanup does not match a cache-strategy typo bug." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel not applied in model_init is different from flash_attention_3 training degeneration." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Different models and failures: Qwen3-VL bbox output vs Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "PatchTSMixer post_init policy issue vs tied-weight key AttributeError." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs still referencing pipeline() vs scheduler kwargs bug; not the same change." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation clarification for model.forward is unrelated to tokenizer loading regression." + }, + { + "left": "issue:41669", + "right": "issue:44908", + "accept": false, + "reason": "Import-star cleanup/perf issue is unrelated to scheduler behavior." + }, + { + "left": "issue:43723", + "right": "issue:44393", + "accept": false, + "reason": "Tokenizer loading failure vs Qwen3-VL bounding-box bug; different subsystems." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Saving Pipeline objects is unrelated to tied-weight key AttributeError." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env command and lazy loading bug are unrelated." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer ReLU discussion vs Gemma4Audio positional encoding; not the same issue." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "Generic bug report does not establish the same root cause as the Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading problem vs scheduler kwargs bug." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import-star usage/performance issue is unrelated to VRAM leak across dataloader threads." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Timestamp calculation bug vs GLM-5 RoPE implementation issue." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Qwen3VL timestamp bug is unrelated to scheduler kwargs handling." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Model.forward docs issue vs Llama 3.2 vision model loading break." + }, + { + "left": "issue:31795", + "right": "issue:37428", + "accept": false, + "reason": "Documentation confusion vs flash_attention_utils import error." + }, + { + "left": "issue:44315", + "right": "issue:45092", + "accept": false, + "reason": "Liger Kernel/model_init issue is different from old checkpoint meta-init incompatibility." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak across threads is unrelated to RoPE implementation." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "Different architectures and failures: GLM-5 RoPE vs Gemma4 audio positional encoding." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 API-setting guidance is unrelated to VRAM leak behavior." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "local_files_only API requests vs timestamp calculation bug; different paths." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Same as above: unrelated documentation issue vs model-loading regression." + }, + { + "left": "issue:31795", + "right": "issue:37428", + "accept": false, + "reason": "No shared underlying bug: docs wording vs import failure." + }, + { + "left": "issue:44315", + "right": "issue:45092", + "accept": false, + "reason": "Both involve loading, but one is Liger Kernel application and the other is checkpoint/meta-init compatibility; not the same bug." + }, + { + "left": "issue:43704", + "right": "issue:45341", + "accept": false, + "reason": "VRAM leak issue is unrelated to testing_utils bug." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": true, + "reason": "Both describe the same v5 tokenizer regression; 43066 is the concrete wrong-decoder-type manifestation of the broader behavior change in 42913." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Scheduler kwargs bug is unrelated to the placeholder 'Unemployment' issue." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "Tied-weight key crash and apply_chat_template tool-call crash are different failures." + }, + { + "left": "issue:44998", + "right": "issue:45341", + "accept": false, + "reason": "Placeholder issue title is unrelated to testing_utils." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "Cache-strategy typo does not match the placeholder issue." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to the placeholder issue." + }, + { + "left": "issue:44556", + "right": "issue:44998", + "accept": false, + "reason": "Checkpoint reload incompatibility is unrelated to the placeholder issue." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Llama4Config missing pad_token_id vs tied-weight key AttributeError; different code paths." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is conservative and grounded in the packet. The only accepted soft pair, issue:42913 / issue:43066, is supported by the titles as the specific v5 tokenizer regression versus the broader tokenizer-behavior report. The remaining pair judgments stay on the cautious side and do not overclaim shared root cause from the available evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 13, + "review_unit_index": 42, + "review_unit_count": 42, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-366", + "nodes": [ + "issue:31515", + "issue:31795", + "issue:33453", + "issue:36296", + "issue:36683", + "issue:38617", + "issue:43298", + "issue:43572", + "issue:44230", + "issue:44623", + "issue:44861", + "issue:45310" + ], + "soft_pairs": [ + "issue:43572|issue:44861", + "issue:31515|issue:38617", + "issue:44230|issue:44623", + "issue:31515|issue:36296", + "issue:31515|issue:31795", + "issue:44230|issue:45310", + "issue:31515|issue:33453", + "issue:31515|issue:36683", + "issue:43298|issue:44230" + ], + "prepared_review_unit_hash": "71ff70509adff5f4d7c4fb99dfc5e35ca9465de20eb5b7ae0110da197c9e055c", + "component_budget": { + "node_count": 366, + "item_count": 366, + "soft_pair_count": 1367, + "serialized_chars": 705321, + "estimated_input_tokens": 176331, + "estimated_eval_tokens": 352918 + }, + "budget": { + "node_count": 12, + "item_count": 12, + "soft_pair_count": 9, + "serialized_chars": 9455, + "estimated_input_tokens": 2364, + "estimated_eval_tokens": 4984 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The soft edges look like false positives: the paired issues cover unrelated docs, loading, config, tokenizer, fp8, and training problems. No duplicate or merge-worthy issue pair stands out.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:45310 is the strongest standalone report: it is specific, reproducible, and names a concrete from_pretrained regression tied to transformers>=5.4.0 and Qwen3.5 MoE.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "Different failures: StableLmConfig pad_token_idx regression vs AttributeError in tied-weight key handling. Not the same bug." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Checkpoint loading performance issue vs ImportError for layer_type_validation. Different code paths and symptoms." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 inference support request for Qwen models vs processor.save_pretrained missing files. Unrelated issues." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "from_pretrained slowdown vs tensor-parallel training bug. Not the same underlying problem." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Slow checkpoint loading vs documentation confusion about input_ids/past_key_values. Different scope entirely." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "Both mention Qwen3.5 MoE, but one is FP8 inference support and the other is a from_pretrained runtime error. Not mergeable as one bug." + }, + { + "left": "issue:31515", + "right": "issue:33453", + "accept": false, + "reason": "Checkpoint loading slowness vs tokenizer loading regression. Separate loading layers and failure modes." + }, + { + "left": "issue:31515", + "right": "issue:36683", + "accept": false, + "reason": "General loading slowdown vs Gemma3Config missing vocab_size. Different model/config bug." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Continuous batching audio-model limitation vs FP8 support for Qwen3.5. Different features and code paths." + } + ] + }, + "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": 13, + "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-26350-2", + "nodes": [ + "issue:26350", + "issue:33260" + ], + "soft_pairs": [ + "issue:26350|issue:33260" + ], + "prepared_review_unit_hash": "3c4d23a14b9ac86b9d8543a753ef90f87db16047fea82ca136ecc9c6d5512e5e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "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 issues are both community-contribution feature requests, but they target different features and code paths: Flash Attention 2 support vs GGUF support. They are not duplicates.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 26350 is the stronger representative of the cluster by discussion activity, but the two issues are not the same request, so there is no true canonical duplicate here.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 26350 has substantially more discussion activity and is a clearer, higher-signal feature request than 33260.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Different requested features for different backend/architecture support; they do not describe 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": 4, + "component_count": 13, + "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-27088-4", + "nodes": [ + "issue:27088", + "issue:43834", + "issue:44365", + "issue:44370" + ], + "soft_pairs": [ + "issue:27088|issue:44370", + "issue:27088|issue:44365", + "issue:27088|issue:43834", + "issue:43834|issue:44370", + "issue:43834|issue:44365", + "issue:44365|issue:44370" + ], + "prepared_review_unit_hash": "3f549b441ca0c48ab82b5271d4b6c621b11270bf9cc00fe7cb5fd1be570a63d9", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4454, + "estimated_input_tokens": 1114, + "estimated_eval_tokens": 2484 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4454, + "estimated_input_tokens": 1114, + "estimated_eval_tokens": 2484 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is dominated by nearly identical i18n translation-request templates, but the only concrete item is Turkish. These look like separate language requests rather than duplicates of the same underlying issue, so the high textual similarity is misleading.", + "confidence": 0.89, + "canonical_issue_reason": "issue #27088 is the only specific, actionable request (Turkish) and is the oldest open item; the others are generic placeholder-style translation issues.", + "canonical_pr_reason": null, + "best_issue_reason": "#27088 is the best representative because it names a real language and can actually be acted on, while the others are template placeholders with no concrete language filled in.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Both are translation-request issues, but they appear to be for different language entries / template instances, not the same concrete change or bug." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "The shared wording is just an i18n template; #27088 is specifically Turkish, so this is not the same underlying issue." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "High textual similarity comes from the template, but there is no evidence they target the same language or same work item." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": false, + "reason": "These are near-identical templates, but without a shared concrete language target they should not be merged as duplicates." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": false, + "reason": "The bodies/titles are template-like and likely represent separate language requests rather than one duplicate issue." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": false, + "reason": "Same template family, but no concrete evidence of the same underlying translation task; treat as separate issues." + } + ] + }, + "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": 13, + "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-36817-13", + "nodes": [ + "issue:36817", + "issue:37015", + "issue:41211", + "issue:42503", + "issue:42738", + "issue:43671", + "issue:43837", + "issue:43909", + "issue:44121", + "issue:44205", + "issue:45027", + "issue:45177", + "issue:45322" + ], + "soft_pairs": [ + "issue:36817|issue:41211", + "issue:36817|issue:43671", + "issue:37015|issue:43837", + "issue:37015|issue:44121", + "issue:37015|issue:43909", + "issue:37015|issue:42503", + "issue:37015|issue:45027", + "issue:41211|issue:43671", + "issue:42503|issue:43837", + "issue:42503|issue:43671", + "issue:42503|issue:42738", + "issue:42503|issue:45027", + "issue:42738|issue:45027", + "issue:43671|issue:43837", + "issue:43671|issue:45027", + "issue:43671|issue:45177", + "issue:43671|issue:44121", + "issue:43671|issue:43909", + "issue:43837|issue:43909", + "issue:43837|issue:45322", + "issue:43837|issue:44121", + "issue:43909|issue:44205", + "issue:44121|issue:45322", + "issue:43837|issue:45027", + "issue:43837|issue:45177" + ], + "prepared_review_unit_hash": "a44b4221c633378924b1ce3284565b95f866975bc4b31b9cd2c4078abdd028df", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 25, + "serialized_chars": 16008, + "estimated_input_tokens": 4002, + "estimated_eval_tokens": 8260 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 25, + "serialized_chars": 16008, + "estimated_input_tokens": 4002, + "estimated_eval_tokens": 8260 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a loose set of model-add requests spanning different families and modalities (text BERT variants, speech ASR/TTS, audio, OCR, and one sparse-transformer proposal). They are not the same underlying bug/change, so none of the soft links should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:37015 ('Add NeoBERT') is the most generic and actively discussed model-add request in the set, so it is the best loose representative of the cluster\u2014but it is not a true duplicate target for the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "Different model requests: EuroBert vs DEIMv2 are unrelated model families and likely different implementation paths." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "Text model config request vs Qwen3-TTS speech support; different modality and code path." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT and Qwen3-ASR are different models for different tasks; not the same change." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "A BERT model addition is unrelated to an OpenAI weight-sparse transformer request." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT (text) and LFM2.5 Audio are different model/support requests." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "Both are model-add requests, but for different concrete model families; not a duplicate change." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS are different models/modalities, so they should not be merged." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are unrelated requests with different tasks and implementation paths." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT and Qwen3-ASR are different concrete model additions, not one bug fix." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT models and Qwen3-TTS address different model families and modalities." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT models vs a broad 'BERT-like models with RoPE' proposal is too loose; not the same concrete change." + }, + { + "left": "issue:42503", + "right": "issue:45027", + "accept": false, + "reason": "ModernVBERT and Voxtral TTS are unrelated model requests." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "A general RoPE/BERT-like feature request is not the same as adding Voxtral TTS support." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS and Qwen3-ASR are both speech-related but require different support paths and different concrete model integrations." + }, + { + "left": "issue:43671", + "right": "issue:43909", + "accept": false, + "reason": "Different audio models; same broad area, but not the same underlying change." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS support is unrelated to the weight-sparse transformer proposal." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Both mention TTS, but they are different models and likely different integration work." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "TTS support and DeepSeek-OCR2 are different modalities and code paths." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Both are audio-related, but Qwen3-ASR and LFM2.5 Audio are different concrete model requests." + }, + { + "left": "issue:43837", + "right": "issue:44205", + "accept": false, + "reason": "ASR support and SAM3-LiteText are unrelated model domains." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR and EUPE are different tasks/models with no shared underlying fix." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "OpenAI weight-sparse transformer and EUPE are unrelated architecture/model requests." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Both are speech-related, but ASR vs TTS support for different models is not the same change." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model integrations." + } + ] + }, + "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": 13, + "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-37934-6", + "nodes": [ + "issue:37934", + "issue:42936", + "issue:43240", + "issue:43668", + "issue:43813", + "issue:44031" + ], + "soft_pairs": [ + "issue:37934|issue:43668", + "issue:42936|issue:43240", + "issue:42936|issue:44031", + "issue:42936|issue:43813", + "issue:42936|issue:43668", + "issue:43240|issue:43668", + "issue:43668|issue:44031", + "issue:43668|issue:43813" + ], + "prepared_review_unit_hash": "f2801e03c578edf992d7bdb1778d3cb2b615c0c168d4268820def1ad304b6b77", + "component_budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6293, + "estimated_input_tokens": 1574, + "estimated_eval_tokens": 3404 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6293, + "estimated_input_tokens": 1574, + "estimated_eval_tokens": 3404 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These issues are only superficially similar by wording; they cover different models, config annotations, loss plumbing, quantization typos, and a tokenizer warning regression, so none of the soft links look like true duplicates.", + "confidence": 0.78, + "canonical_issue_reason": "No true canonical issue: the set spans distinct bugs and code paths rather than one underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "43240 is the best single anchor if one must be chosen: it is a concrete bug report with the most apparent cross-reference activity.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Both mention normalization, but one is about Llama4TextL2Norm behavior and the other is a ModernBERT config type hint; different models and fixes." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Both concern training/loss plumbing, but Mask2Former ignore_value handling and kwargs forwarding to loss are different bugs in different paths." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Mask2Former ignore_value usage and tokenizer regex warnings are unrelated subsystems and symptoms." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "Loss-argument handling in Mask2Former is unrelated to a quantization typo string." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "A model loss-plumbing bug is not the same as an incorrect config type hint; different scope and remediation." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Forwarding kwargs to a loss function and correcting a config annotation are separate issues with no shared code path." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "ModernBERTConfig type hints and tokenizer regex warnings are unrelated bugs." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "A type-hint typo in config is unrelated to a quantization-operation typo." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 7, + "component_count": 13, + "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": "cc88fe5f819a9686e46d27ffff34b0e5fd1258db9af892bba03b07313ca98249", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "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: one requests block decoding, the other requests DeepSeek v3.2 support. They touch different feature areas and describe different needs.", + "confidence": 0.95, + "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: block decoding vs model/version support. No evidence they refer to 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": 8, + "component_count": 13, + "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": "fa3d23b0ceb21d8993c92f8de769fda26035581fd4a814ac8c3a8603df6abeb2", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues are about different Trainer concerns: one requests better customizability around `remove_unused_column`, while the other asks FLOPS estimation to account for custom trainers. They are related to Trainer internals but do not describe the same bug or change, so they should not be deduplicated.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "Different underlying requests: customizability of `remove_unused_column` vs FLOPS estimation for custom trainers. Shared Trainer context is too broad to treat as a duplicate." + } + ] + }, + "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": 13, + "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": "675303e5bfe517616d1b7b783857ea61ca0888674308221762b0eb745aa66918", + "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 two issues are about different Qwen-related feature gaps: one requests model registration for Qwen3-Omni on AutoModel/AutoModelForConditionalGeneration, while the other requests sequence-classification support for Qwen3.5. They are not duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical duplicate is evident; if one must be picked, issue #43684 is the broader API-registration request, but it is still a different feature from #44405.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #43684 is the more general representative of a model-registration gap, but the cluster is not truly cohesive enough to treat either issue as a duplicate of the other.", + "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 versus Qwen3.5 sequence-classification support. They affect different model/config paths and 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": 10, + "component_count": 13, + "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": "d67c337eaeea102fe3702bda9c728c2b2ff119e2b77e8842a50bbbe9b7db1854", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Three issues share the identical vague title \"huggingface\" and look duplicate-like; the \"Fett\" issue appears separate. I\u2019d cluster the three huggingface reports together, with 43845 as the canonical duplicate target and 43859 as the most informative representative.", + "confidence": 0.74, + "canonical_issue_reason": "43845 is the earliest of the repeated \"huggingface\" reports and cleanly represents the duplicate-like subgroup.", + "canonical_pr_reason": null, + "best_issue_reason": "43859 has the most discussion/activity and likely the richest context, so it is the best triage-facing representative.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different titles and no concrete shared bug details; the weak similarity is not enough to treat them as the same issue." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Same as above: \"Fett\" does not provide evidence of the same underlying problem as the huggingface reports." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "No substantive overlap beyond low soft similarity; insufficient evidence they describe the same bug." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": true, + "reason": "Same title and very high similarity; they appear to be duplicate reports of the same issue." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": true, + "reason": "Same title and strong similarity suggest the same underlying report, despite differing body length." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": true, + "reason": "Same title and strong similarity; likely duplicate reports of the same issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The clustering is grounded in the packet and stays appropriately cautious: 43845/43846/43859 share the identical title and the strongest soft-similarity scores, while 43835 has a different title and only weak similarity. The chosen canonical/representative issue assignments are conservative enough." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 13, + "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-44371-5", + "nodes": [ + "issue:44371", + "issue:44418", + "issue:44572", + "issue:44573", + "issue:44937" + ], + "soft_pairs": [ + "issue:44371|issue:44572", + "issue:44371|issue:44573", + "issue:44371|issue:44418", + "issue:44371|issue:44937", + "issue:44418|issue:44572", + "issue:44418|issue:44573", + "issue:44572|issue:44573", + "issue:44572|issue:44937", + "issue:44573|issue:44937" + ], + "prepared_review_unit_hash": "0f1beca16484909016a6b8dfd2eaec788144cc722c5da7d7ff132f5e30fcde53", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5794, + "estimated_input_tokens": 1449, + "estimated_eval_tokens": 3154 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5794, + "estimated_input_tokens": 1449, + "estimated_eval_tokens": 3154 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are mostly unrelated low-information/spam issues, with one generic documentation request. There\u2019s no solid evidence of a shared underlying bug or change across the set, so this cluster should not be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 44418 is the only non-spam-sounding item, so it is the least-bad anchor if a canonical issue must be chosen, but it still does not appear related to the others.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44418 is the most substantive of the group and the only one with a real topic in the title; the rest are spam-like placeholders.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both are spam-like placeholders; no concrete shared problem is described." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Both are spam-like placeholders; no evidence they describe the same issue." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is spam, the other is a generic documentation suggestion; different intent and content." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder versus a specific quoted-name topic; not the same underlying issue." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Generic documentation request and spam placeholder are unrelated." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "Generic documentation request and spam placeholder are unrelated." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Both are spam-like placeholders, but there is no concrete bug/change to unify." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder versus a specific quoted-name topic; no shared underlying report." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder versus a specific quoted-name topic; no shared underlying report." + } + ] + }, + "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": 13, + "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": "f7bfb66cdd8102dca285d7dedd8a82997bac8fd3aa2e83b044bcf1ab11d2b5c9", + "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 two closed issues are strong near-duplicates based on identical titles and very similar bodies, but the packet does not prove they are the same underlying report. #45145 is the better canonical representative because it was created first and has an inbound reference.", + "confidence": 0.78, + "canonical_issue_reason": "#45145 is earlier and already has an inbound reference, so it is the best canonical issue to retain.", + "canonical_pr_reason": null, + "best_issue_reason": "#45145 is the strongest issue to keep as canonical because it was created first and has more linkage than #45151.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": false, + "reason": "Identical titles and soft-similarity make this a likely duplicate, but the evidence does not establish identical body content or confirm the same underlying issue. Conservative triage should not mark it as an accepted duplicate on this packet alone." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The canonical-choice rationale is partly grounded (45145 is earlier and has one inbound reference), but the soft-edge acceptance is too strong for the evidence. The packet only shows identical titles and a soft-similarity signal; it does not establish that the bodies are the same or that these are definitely duplicates. Also, \"slightly more linkage/activity\" is not supported as stated because 45151 has one discussion activity while 45145 does not. A more conservative verdict would be to mark the pair as a likely duplicate rather than an accepted duplicate." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 13, + "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": "pull_request", + "component_index": 1, + "component_count": 24, + "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-39895-11", + "nodes": [ + "pull_request:39895", + "pull_request:42668", + "pull_request:43251", + "pull_request:43254", + "pull_request:43424", + "pull_request:43823", + "pull_request:43995", + "pull_request:44827", + "pull_request:45170", + "pull_request:45213", + "pull_request:45435" + ], + "soft_pairs": [ + "pull_request:39895|pull_request:43424", + "pull_request:42668|pull_request:43823", + "pull_request:43251|pull_request:43823", + "pull_request:43251|pull_request:44827", + "pull_request:43424|pull_request:44827", + "pull_request:43424|pull_request:45435", + "pull_request:43424|pull_request:43823", + "pull_request:43823|pull_request:45435", + "pull_request:43823|pull_request:43995", + "pull_request:43823|pull_request:44827", + "pull_request:43995|pull_request:45435", + "pull_request:43995|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:45213", + "pull_request:45213|pull_request:45435", + "pull_request:43823|pull_request:45213", + "pull_request:44827|pull_request:45435", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "39ec3bb4ea751e1f7b383e73ec3002f0a5c4b0afaaad2bba16c7a50c33869f3d", + "component_budget": { + "node_count": 11, + "item_count": 11, + "soft_pair_count": 19, + "serialized_chars": 31458, + "estimated_input_tokens": 7865, + "estimated_eval_tokens": 15986 + }, + "budget": { + "node_count": 11, + "item_count": 11, + "soft_pair_count": 19, + "serialized_chars": 31458, + "estimated_input_tokens": 7865, + "estimated_eval_tokens": 15986 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This set is mostly unrelated PRs spanning model additions, refactors, docs, and tests. The only true duplicate-like pair is the two `fixed_cross_entropy` fixes for issue #43240, which target the same function and underlying missing kwargs support.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43251 is the clearest canonical representative: it explicitly targets issue #43240 and fixes `fixed_cross_entropy` to forward missing `cross_entropy` kwargs in the same code path as #43254.", + "best_issue_reason": null, + "best_pr_reason": "PR #43251 is the best representative overall because it is tightly scoped, issue-linked, and addresses the same concrete loss-function bug as #43254.", + "soft_edge_verdicts": [ + { + "left": "pull_request:39895", + "right": "pull_request:43424", + "accept": false, + "reason": "Unrelated changes: new VideoPrism model/docs versus an ExecuTorch dynamic-shape test." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Different workstreams: processor auto-loading robustness versus adding MobileLLM." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Loss helper kwargs fix is unrelated to adding MobileLLM." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Same broad library area, but one is a loss-function fix and the other is Mistral4 test/model work." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "ExecuTorch export test and Mistral4 tests are different concrete changes." + }, + { + "left": "pull_request:43424", + "right": "pull_request:45435", + "accept": false, + "reason": "Dynamic-shape export test versus Whisper tokenizer bugfix; no shared underlying change." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Different features: ExecuTorch testing versus new MobileLLM model support." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45435", + "accept": false, + "reason": "MobileLLM model addition and Whisper tokenizer fix are unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "MobileLLM addition versus Falcon output-interface refactor; not the same bug/change." + }, + { + "left": "pull_request:43823", + "right": "pull_request:44827", + "accept": false, + "reason": "New MobileLLM model work is unrelated to Mistral4 test fixes." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45435", + "accept": false, + "reason": "Falcon refactor and Whisper tokenizer fix touch different code paths and problems." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon model refactor is unrelated to the model-creation skill draft PR." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon standardized output collection refactor and Mistral4 tests are not the same underlying change." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test/model fixes versus a broad `layrnorm` rename across CLIP-like models; different bugs." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 test fixes and the model-creation skill PR are unrelated." + }, + { + "left": "pull_request:45213", + "right": "pull_request:45435", + "accept": false, + "reason": "A meta skill/documentation PR is not the same as a Whisper tokenizer bugfix." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "MobileLLM model addition and the model-creation skill draft do not fix the same concrete code-path problem." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45435", + "accept": false, + "reason": "Mistral4 test fixes and Whisper tokenizer bugfix are unrelated." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Both PRs fix the same `fixed_cross_entropy` kwargs forwarding bug in `loss_utils.py` for issue #43240, and could plausibly be merged into one change." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The evaluation is well-grounded in the packet. The summary stays conservative, and the only accepted soft-edge pair, PR #43251 and PR #43254, is supported by the shared `loss_utils.py` change and the same issue target (#43240). The other pairwise rejections are also appropriately cautious and do not overclaim mergeability." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 2, + "component_count": 24, + "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-40546-2", + "nodes": [ + "pull_request:40546", + "pull_request:42781" + ], + "soft_pairs": [ + "pull_request:40546|pull_request:42781" + ], + "prepared_review_unit_hash": "dc199947d14aa90df5aa9dda5c5859c2b9fa98c59bca603d9e9847506b0bf208", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7225, + "estimated_input_tokens": 1807, + "estimated_eval_tokens": 3870 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7225, + "estimated_input_tokens": 1807, + "estimated_eval_tokens": 3870 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related VibeVoice PRs, but they are not duplicates: #40546 adds the initial VibeVoice implementation, while #42781 adds a broader realtime variant plus an acoustic tokenizer and additional generation/model changes. Overlap is mostly shared model scaffolding and docs, not the same concrete fix/change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "#42781 is the broader and later VibeVoice PR, covering the realtime path and acoustic tokenizer in addition to the base model, so it is the better representative if one PR must be chosen for the cluster.", + "best_issue_reason": null, + "best_pr_reason": "#42781 is the most complete VibeVoice PR in this set and subsumes more of the related implementation surface, making it the better anchor candidate for triage.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Reject: they share VibeVoice infrastructure files, but the PRs target different changesets\u2014initial VibeVoice support vs. realtime + acoustic tokenizer expansion. They do not look like the same underlying bug or a single 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": 3, + "component_count": 24, + "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": "fea9054a212e574d7670f8c791e82c42a65e33d59924b7cbcf1105f30cfd44c2", + "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": "The cluster contains two unrelated pull requests: one adds TimesFM covariate-based forecasting, the other adds DINOv3 image classification support. They do not fix the same bug or implement the same change, so they should not be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR fits this cluster: the two PRs target different model families and different features, with no shared underlying code-path problem.", + "best_issue_reason": null, + "best_pr_reason": "PR #40755 is the broader, more feature-complete TimesFM change, but it is not a representative duplicate of the DINOv3 classification PR; overall, neither PR is a good cluster-wide canonical.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different models and features: TimesFM covariate forecasting vs. DINOv3 image classification support. They are not the same underlying change and would not 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": 4, + "component_count": 24, + "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-41356-2", + "nodes": [ + "pull_request:41356", + "pull_request:44339" + ], + "soft_pairs": [ + "pull_request:41356|pull_request:44339" + ], + "prepared_review_unit_hash": "a3c962d2540926578e2da616c4d35ea269f16dd241073ecbf1aac9c482085f67", + "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": "Both pull requests are about adding DEIMv2 to Transformers. PR 44339 is the more complete, later integration and appears to supersede the earlier basic implementation in 41356.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "44339 is the more complete DEIMv2 integration and supersets 41356 with broader model wiring, auto-mapping updates, conversion tooling, loss support, and config checks.", + "best_issue_reason": null, + "best_pr_reason": "44339 is the strongest representative of the cluster because it covers the full DEIMv2 addition rather than only the initial model/image-processor scaffold.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41356", + "right": "pull_request:44339", + "accept": true, + "reason": "Both PRs implement the same underlying change: adding DEIMv2 support. 44339 is effectively a superset of 41356, so they are close enough to treat as the same feature change." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs target issue 41211 and add DEIMv2, with 44339 clearly extending the earlier scaffold in 41356 via auto-mapping, conversion, loss, and config-related changes. The soft-edge verdict is conservative enough because the two changes look like mergeable iterations of the same feature work." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 24, + "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-41797-2", + "nodes": [ + "pull_request:41797", + "pull_request:45186" + ], + "soft_pairs": [ + "pull_request:41797|pull_request:45186" + ], + "prepared_review_unit_hash": "7443f734b60321912bc7423de9cf15dee6ff9d8a12b1d8eeed23ba618e973839", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6721, + "estimated_input_tokens": 1681, + "estimated_eval_tokens": 3618 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6721, + "estimated_input_tokens": 1681, + "estimated_eval_tokens": 3618 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two separate model-introduction PRs (DeepSeek OCR vs Isaac). They share the same Transformers integration pattern and some registry/doc files, but they target different models and different code paths, so they are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the PRs implement different models and cannot be collapsed into one duplicate change.", + "best_issue_reason": null, + "best_pr_reason": "No single best PR for deduping here; neither PR subsumes the other because they add unrelated model integrations.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41797", + "right": "pull_request:45186", + "accept": false, + "reason": "Reject: both are new-model integration PRs, but for different models (DeepSeek OCR vs Isaac) with distinct configs, models, processors, tests, and docs. Shared registry/docs scaffolding does not indicate 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": "pull_request", + "component_index": 6, + "component_count": 24, + "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-42166-2", + "nodes": [ + "pull_request:42166", + "pull_request:45426" + ], + "soft_pairs": [ + "pull_request:42166|pull_request:45426" + ], + "prepared_review_unit_hash": "33b1ba550b4994f8e2a9729e5c43b0b60e3a40ae4e20035663a4a4352064cb3a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5408, + "estimated_input_tokens": 1352, + "estimated_eval_tokens": 2960 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5408, + "estimated_input_tokens": 1352, + "estimated_eval_tokens": 2960 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two pull requests are unrelated model-addition PRs: one adds InternVLFlash, the other adds AXK1. They share only generic auto-registration plumbing and are not the same underlying change.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "No strong canonical PR exists here; if one must be chosen as the cluster representative, PR #42166 is slightly more substantial and established (more files changed and higher review activity), but it is not a duplicate of PR #45426.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42166", + "right": "pull_request:45426", + "accept": false, + "reason": "Reject: these are different model-feature PRs for different architectures (InternVLFlash vs AXK1). Shared auto-registration files are incidental and do not indicate the same concrete code-path 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": 7, + "component_count": 24, + "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-42227-5", + "nodes": [ + "pull_request:42227", + "pull_request:43291", + "pull_request:43665", + "pull_request:44815", + "pull_request:45574" + ], + "soft_pairs": [ + "pull_request:42227|pull_request:43291", + "pull_request:42227|pull_request:43665", + "pull_request:42227|pull_request:44815", + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:45574", + "pull_request:44815|pull_request:45574" + ], + "prepared_review_unit_hash": "e4d67011c963b429746bb6e144ec1c0e6cf89f2279da7af0cbff645a5afddd40", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 13477, + "estimated_input_tokens": 3370, + "estimated_eval_tokens": 6996 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 13477, + "estimated_input_tokens": 3370, + "estimated_eval_tokens": 6996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All candidates are pull requests; none are strong duplicate/merge candidates with each other. The Whisper PR and the Whisper tests/tokenization PR touch related area but fix different behaviors, while the remaining PRs are clearly unrelated or broad cosmetic/test-only changes.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #42227 is the clearest focused fix: it directly targets an explicit issue, changes one concrete ASR code path, and has a narrowly scoped diff for Whisper return_language handling.", + "best_issue_reason": null, + "best_pr_reason": "PR #42227 is the best standalone candidate because it has a specific bug report link, a small targeted change, and a clear functional purpose compared with the broader or test-only alternatives.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42227", + "right": "pull_request:43291", + "accept": false, + "reason": "Both are Whisper-related, but #42227 fixes ASR pipeline return_language propagation while #43291 changes tokenizer decode/test behavior; different code paths and different bugs." + }, + { + "left": "pull_request:42227", + "right": "pull_request:43665", + "accept": false, + "reason": "#42227 is a Whisper runtime fix; #43665 is a broad test-only edit across CLIP/ViT with no shared bug or concrete code-path overlap." + }, + { + "left": "pull_request:42227", + "right": "pull_request:44815", + "accept": false, + "reason": "#42227 addresses Whisper return_language in the pipeline; #44815 is a larger FP8/dequant loading change. They are unrelated fixes." + }, + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "#43291 is a Whisper tokenizer behavior change; #43665 only alters tests in unrelated vision models." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "#43291 concerns Whisper decode logic, while #44815 concerns FP8 dequant/loading infrastructure. Different subsystems and bugs." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "#43665 is test noise in CLIP/ViT; #44815 is a substantive quantization/loading fix. No shared underlying change." + }, + { + "left": "pull_request:43665", + "right": "pull_request:45574", + "accept": false, + "reason": "#43665 is unrelated vision test edits; #45574 adds a new model/docs/registry support. Different changes entirely." + }, + { + "left": "pull_request:44815", + "right": "pull_request:45574", + "accept": false, + "reason": "#44815 is an FP8/dequant fix; #45574 is new model onboarding plus docs and registry updates. 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": "pull_request", + "component_index": 8, + "component_count": 24, + "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": "84970cd01fbd704304ff489ebaab2dd47fc3cda0aba17a576072b0f522501473", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5934, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5934, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs both update GitHub Actions dependencies, but they target different action families and different maintenance goals. They are not the same underlying change and should not be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the broader, more central maintenance change: it updates core GitHub Actions used across many workflows for Node 24 compatibility.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the better representative PR for the cluster because it has the wider impact and clearer compatibility-driven scope.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Both are workflow dependency upgrades, but 42975 updates actions/checkout, actions/cache, and upload-artifact for Node 24 compatibility, while 42976 updates docker/build-push-action and related workflow versions. They touch overlapping CI files but fix different action-version maintenance issues, 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": 9, + "component_count": 24, + "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:43451", + "pull_request:45490" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:43451", + "pull_request:42978|pull_request:45490", + "pull_request:43448|pull_request:43451", + "pull_request:43448|pull_request:45490", + "pull_request:43451|pull_request:45490" + ], + "prepared_review_unit_hash": "0ae72507573ba71181825e6c0fc141d452861230c5bdf97f4275d46b8d8a1c13", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14136, + "estimated_input_tokens": 3534, + "estimated_eval_tokens": 7324 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14136, + "estimated_input_tokens": 3534, + "estimated_eval_tokens": 7324 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All four items are separate model-addition PRs that only overlap on Transformers boilerplate (docs, auto mappings, tests). They do not describe the same underlying change, so no soft pair should be merged.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR stands out as a duplicate target; if forced to pick one representative, 43451 is the most substantial/complete, but it is still a distinct Molmo2 addition rather than a duplicate of the others.", + "best_issue_reason": null, + "best_pr_reason": "43451 is the strongest standalone PR in the set by scope and review activity, but it is not duplicate-equivalent to the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "ViT NEPA vs Molmo: different model families and different implementation files, so this is not the same code change." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "ViT NEPA vs Molmo2: unrelated model additions; shared auto/docs scaffolding is too generic to indicate a duplicate." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "ViT NEPA vs CTSM are distinct model integrations with different configs/modeling paths; not the same underlying feature." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Molmo vs Molmo2 are related by naming but they add different models and codepaths, not one mergeable fix/change." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo and CTSM are unrelated model additions; overlap is limited to shared registration/docs updates." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo2 vs CTSM are different model implementations with no concrete shared bug or change to merge together." + } + ] + }, + "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": 24, + "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:45476" + ], + "soft_pairs": [ + "pull_request:43488|pull_request:43532", + "pull_request:43488|pull_request:45476", + "pull_request:43532|pull_request:45476" + ], + "prepared_review_unit_hash": "37be47827d08eb8738a1bb676fff354d941e4ea415dea2086a19c2a47569ad4b", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7469, + "estimated_input_tokens": 1868, + "estimated_eval_tokens": 3992 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7469, + "estimated_input_tokens": 1868, + "estimated_eval_tokens": 3992 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three unrelated '[don't merge]' demo PRs with only incidental overlap in a shared test file; they do not look like duplicates or the same underlying fix.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43532 is the best representative of the cluster because it is the broadest example PR, with the most files and workflow-related changes, while the others are narrower demo changes.", + "best_issue_reason": null, + "best_pr_reason": "PR #43532 is the most representative open PR here: it has the largest surface area and the clearest 'show diff' purpose, making it the best cluster anchor despite not matching the others semantically.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "Shared only the vit test file; one PR is a setup/version/dependency tweak, the other is workflow/testing-utils demo changes. Different code paths and purposes." + }, + { + "left": "pull_request:43488", + "right": "pull_request:45476", + "accept": false, + "reason": "Only incidental overlap on a test file. #43488 is packaging/version/dependency related; #45476 is a CI-workflow call plus unrelated tokenization/test edits." + }, + { + "left": "pull_request:43532", + "right": "pull_request:45476", + "accept": false, + "reason": "Both touch CI/test areas, but the concrete edits are different: workflow/testing-utils versus a workflow-caller plus clip/chinese_clip test and tokenizer stubs. 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": 11, + "component_count": 24, + "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": "50d5c410d340e7f730b38bad6591181a007e3e897b1946062a59703f75cc1a53", + "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": "These are related but distinct PRs: one adds a Promptable Concept Segmentation pipeline, the other adds a Promptable Visual Segmentation pipeline. The overlap is mostly shared registration/docs scaffolding; the actual targets, pipeline names, and model/test files differ, so they should not be deduplicated.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR 43613 is the stronger representative of the cluster because it includes broader concrete implementation and test coverage across multiple segmentation model families, making it the more complete standalone change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Different pipeline features with different task names and model support; shared docs/registration code is generic scaffolding, 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": 12, + "component_count": 24, + "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": "08af0f941814e2205b1c5d106a540c0fd98b05df8352d43c45273b45a800f28d", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4292, + "estimated_input_tokens": 1073, + "estimated_eval_tokens": 2402 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4292, + "estimated_input_tokens": 1073, + "estimated_eval_tokens": 2402 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests add small extensibility hooks to `Trainer`, but they address different code paths: gradient-accumulation loss scaling vs. dataset column/signature filtering. They are not duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different underlying changes in `trainer.py`: one alters training-step loss scaling, the other extracts signature-column inspection for column filtering. They are related only at the subsystem level, not 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": 13, + "component_count": 24, + "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-43656-3", + "nodes": [ + "pull_request:43656", + "pull_request:43836", + "pull_request:43842" + ], + "soft_pairs": [ + "pull_request:43656|pull_request:43842", + "pull_request:43836|pull_request:43842" + ], + "prepared_review_unit_hash": "7a91a46c8f5292986da50d6dd5e1711474a63c04818142f29065b4bbe4508c3b", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5993, + "estimated_input_tokens": 1499, + "estimated_eval_tokens": 3254 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5993, + "estimated_input_tokens": 1499, + "estimated_eval_tokens": 3254 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three PRs target the same TypeAdapter annotation/import problem in `src/transformers/cli/serve.py`. PR 43842 is the most focused fix; 43656 and 43836 look like earlier or broader variants of the same underlying NameError issue.", + "confidence": 0.86, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43842 is the cleanest and most specific fix: it directly addresses the `TypeAdapter` NameError when pydantic is absent, with minimal surface area.", + "best_issue_reason": null, + "best_pr_reason": "PR 43842 is the best representative because it targets the exact runtime failure with the smallest, most explicit change set.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": true, + "reason": "Both appear to fix the same `TypeAdapter` annotation NameError in `cli/serve.py`; 43656 does it via future annotations/type-hint cleanup, while 43842 uses a more explicit TYPE_CHECKING import/quoted annotation. Same underlying bug, mergeable into one fix." + }, + { + "left": "pull_request:43836", + "right": "pull_request:43842", + "accept": true, + "reason": "Both are about the same `TypeAdapter` NameError in the CLI serve path and touch the same code locations. 43836 is broader, but the concrete bug being fixed is the same and could plausibly be handled in one PR." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and both soft-edge accepts are conservative enough: 43836 and 43842 clearly overlap on the same serve.py TypeAdapter fix, and 43656 is at least plausibly the same runtime-annotation bug in the same file. No evidence of an overstatement that would justify rejecting." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 24, + "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": 15, + "component_count": 24, + "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-43838-3", + "nodes": [ + "pull_request:43838", + "pull_request:43973", + "pull_request:45283" + ], + "soft_pairs": [ + "pull_request:43838|pull_request:45283", + "pull_request:43973|pull_request:45283" + ], + "prepared_review_unit_hash": "a577666818cc64ae8e846b3dd2c6891718952d9bcd7f6c422d518f83a0ae1523", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 8697, + "estimated_input_tokens": 2175, + "estimated_eval_tokens": 4606 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 8697, + "estimated_input_tokens": 2175, + "estimated_eval_tokens": 4606 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are related only by broad Qwen/audio metadata, not by the same underlying change. 43838 is the strongest standalone PR here; 45283 is a separate GGUF-loading mapping update, and 43973 is another distinct audio model addition. None of the soft pairs should be merged as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43838 is the most mature and self-contained candidate: it has an explicit issue target, much more discussion/review activity, and a full implementation/test/docs footprint.", + "best_issue_reason": null, + "best_pr_reason": "PR 43838 is the best representative for the cluster because it is the most complete, most reviewed, and most clearly scoped model-support addition; the other PRs are separate feature tracks.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43838", + "right": "pull_request:45283", + "accept": false, + "reason": "43838 adds Qwen3-ASR model support and docs; 45283 only adds Qwen3.5 GGUF metadata/mapping support. They touch different code paths and could not plausibly be merged into one PR." + }, + { + "left": "pull_request:43973", + "right": "pull_request:45283", + "accept": false, + "reason": "43973 adds Lfm2Audio model support, while 45283 is a GGUF-loading integration change for Qwen3.5. Same broad audio/modeling area, but 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": "pull_request", + "component_index": 16, + "component_count": 24, + "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:44066|pull_request:44068", + "pull_request:44018|pull_request:44066", + "pull_request:44019|pull_request:44722", + "pull_request:44086|pull_request:44722", + "pull_request:44019|pull_request:44085", + "pull_request:44018|pull_request:44019", + "pull_request:44019|pull_request:44071", + "pull_request:44068|pull_request:44116", + "pull_request:44068|pull_request:44722", + "pull_request:44071|pull_request:44722", + "pull_request:44018|pull_request:44116", + "pull_request:44018|pull_request:44086", + "pull_request:44018|pull_request:44085", + "pull_request:44068|pull_request:44086", + "pull_request:44071|pull_request:44085", + "pull_request:44059|pull_request:44086", + "pull_request:44059|pull_request:44722", + "pull_request:44085|pull_request:44086", + "pull_request:44098|pull_request:44101", + "pull_request:44018|pull_request:44071", + "pull_request:44017|pull_request:44722", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44073", + "pull_request:44068|pull_request:44071", + "pull_request:44071|pull_request:44086", + "pull_request:43997|pull_request:44073", + "pull_request:44059|pull_request:44071", + "pull_request:44017|pull_request:44071", + "pull_request:44073|pull_request:44101", + "pull_request:44059|pull_request:44068", + "pull_request:44018|pull_request:44059", + "pull_request:44056|pull_request:44101", + "pull_request:44073|pull_request:44161", + "pull_request:44101|pull_request:44161", + "pull_request:44017|pull_request:44018", + "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:43997|pull_request:44071", + "pull_request:44010|pull_request:44071", + "pull_request:44010|pull_request:44073", + "pull_request:44018|pull_request:44129", + "pull_request:43997|pull_request:44074", + "pull_request:43997|pull_request:44129", + "pull_request:44010|pull_request:44018", + "pull_request:44071|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": "675923cc03cb926bfc0681b295d1d1d5f3958ccfc2817c163fee1432f6876ce0", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84505, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 + }, + "budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 72, + "serialized_chars": 76938, + "estimated_input_tokens": 19235, + "estimated_eval_tokens": 38726 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a family of model-specific refactors to standardized output tracing/decorator-based return handling. Most PRs are related only by the umbrella output-tracing effort, not by the same concrete code change. The only strong near-duplicate is the GPT-J/CodeGen refactor pair.", + "confidence": 0.72, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44722 is the strongest representative: it is an explicit output-tracing refactor with the shared 43979 target, touches the concrete GPT-J/CodeGen code path, and reads like the most complete version of that specific change.", + "best_issue_reason": null, + "best_pr_reason": "PR 44722 best captures the cluster\u2019s core change pattern and is the clearest anchor for the GPT-J/CodeGen duplicate candidate set.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Same general output-tracing theme, but different model scope; not the same concrete change." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Both are standardized output refactors, but they target different models and code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": false, + "reason": "Very similar theme, but 44066 also changes CodeGen; scope is not clearly the same PR." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "Both are refactors for output tracing, but one is ResNet/RegNet and the other is EfficientNet; not duplicates." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "Different model families and different forward paths; umbrella issue match is not enough." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "Unrelated model families; only share the broad output-tracing migration theme." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "Different backbones and different implementation details; not the same underlying bug/change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "Both are decorator refactors, but they affect different models and code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "Same broad refactor theme, but MPT and GPT-J are distinct code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": false, + "reason": "Related output-tracing refactors, but GPT-Neo and GPT-J are not the same change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen versus GPT-Neo; shared refactor style is not enough." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "Different models and different forward implementations." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet versus GPT-J/CodeGen; only the umbrella refactor pattern matches." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are unrelated model-specific changes." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet refactor and GPT-J refactor are distinct code changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are different implementations; not a duplicate." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "Different backbone families; same migration style only." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo versus Flaubert; not the same bug/change." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo and GPT-J/CodeGen are separate refactors despite similar goals." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "MPT and GPT-J/CodeGen are different model paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "Different models and different output-handling implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated; only the output-tracing theme is shared." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "Different models and different scopes." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are distinct code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44086", + "accept": false, + "reason": "MPT and MGP-STR are unrelated refactors." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44073", + "accept": false, + "reason": "RegNet and VisualBert are different models; not the same change." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 and MPT are separate implementations; same refactor theme only." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J output tracing versus MGP-STR output tracing; different code paths." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert are unrelated model-specific refactors." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44161", + "accept": false, + "reason": "VisualBert and LongT5 are different models; not duplicates." + }, + { + "left": "pull_request:44101", + "right": "pull_request:44161", + "accept": false, + "reason": "Flaubert and LongT5 target different architectures and return paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44018", + "accept": false, + "reason": "SegFormer and GPT-Neo are unrelated changes." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are different models; no shared concrete bug." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are separate output-tracing refactors." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert are different code paths." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet and LongT5 are unrelated beyond the generic refactor theme." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet and GPT-Neo are different model implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet and LongT5 are unrelated." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44161", + "accept": false, + "reason": "TextNet and LongT5 do not share the same underlying change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44071", + "accept": false, + "reason": "RegNet and MPT are different backbones; same umbrella issue is insufficient." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are separate model-specific refactors." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are different architectures and code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-Neo and SpeechT5 are unrelated model families." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44129", + "accept": false, + "reason": "RegNet and SpeechT5 are not the same underlying change." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "SqueezeBert and GPT-Neo are separate refactors." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 have different forward/output paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "CVT/FNet and LongT5 are unrelated beyond the refactor pattern." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "FNet/CVT and SegFormer are different model-specific changes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBert and DeBERTa-v2 are not the same code-path fix." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTa-v2 and TextNet are separate output-tracing refactors." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor on the same code path; 44722 looks like a later rebased/expanded version of the same change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": false, + "reason": "Very similar topic, but 44085 is GPT-J-only while 44722 also includes CodeGen; not clearly the same exact change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J output tracing versus SpeechT5 output tracing are unrelated." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and ResNet/RegNet are different model families; no duplicate evidence." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "FNet/CVT and SpeechT5 are unrelated." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTa-v2 and SpeechT5 are separate model-specific refactors." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa-v2 and VITS are different code paths and architectures." + }, + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "FocalNet and ImageGPT are unrelated models; same refactor theme only." + } + ] + }, + "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": 24, + "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": "ef6a9f020cda65bc8a8cccc2827232d9d7a6745c8804f9843d2c59f8c3a80b42", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84505, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 5, + "serialized_chars": 13067, + "estimated_input_tokens": 3267, + "estimated_eval_tokens": 6790 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These PRs are all separate model-by-model refactors for standardized output tracing/capture, not duplicates of the same code change. They share the same migration theme but touch different model implementations and code paths.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical duplicate PR stands out; the closest representative is 44073 because it cleanly wires in capture_outputs and _can_record_outputs for one model, but the other PRs are still distinct model-specific changes.", + "best_issue_reason": null, + "best_pr_reason": "44073 is the best representative of the cluster\u2019s theme: a focused, self-contained standardized output-tracing refactor with explicit model output-recording setup.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "Both are output-tracing refactors, but for different models (VisualBert vs TextNet) with different code paths; same migration pattern, not the same underlying change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "Both add standardized output capture, but they target different models (MPNet vs VisualBert) and different model internals; not mergeable as one duplicate PR." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "Same refactor family, but distinct model implementations (MPNet vs TextNet) and separate forward/output wiring; not the same concrete bug or change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are unrelated model code paths; these are parallel refactors, not duplicate fixes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "SqueezeBert output-collection refactor and SpeechT5 output-tracing refactor are different model-specific changes despite similar terminology." + } + ] + }, + "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": 24, + "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: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": "e39351b5510747e009cf6432229fb40128fe942d1722c5dc313c88a77262b824", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 39, + "serialized_chars": 39780, + "estimated_input_tokens": 9945, + "estimated_eval_tokens": 20146 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 39, + "serialized_chars": 39780, + "estimated_input_tokens": 9945, + "estimated_eval_tokens": 20146 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a family of independent PRs that all refactor output tracing / `can_return_tuple` handling across different model implementations. They share the same theme, but each targets a different concrete code path, so they should not be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No issues are present; among the PRs, #44003 is the best representative because it touches two closely related Mamba implementations and shows the shared output-capturing refactor most clearly.", + "best_issue_reason": null, + "best_pr_reason": "#44003 is the strongest representative of the cluster: it is a more substantial refactor, spans related model files, and best exemplifies the shared output-tracing change pattern.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Same refactor theme, but different models (`timm_backbone` vs `upernet`) and unrelated code paths." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Both are output-tracing refactors, but they affect different model families (`timm_backbone` vs `vision_text_dual_encoder`)." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Shared output-tracing cleanup, but `timm_backbone` and `univnet` are unrelated implementations." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model stacks (`timm_backbone` vs Mamba/FalconMamba); same pattern, not the same bug/change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Both refactor tracing, but `timm_backbone` and `rwkv` are separate code paths with no shared concrete fix." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "Similar tracing refactor, but `mobilenet_v1` and `univnet` are different models and changes are not mergeable as one PR." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "`mobilenet_v1` and `upernet` share a theme only; they fix different model-specific forward paths." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "Both are tracing refactors, but they apply to different depth-related models and distinct code paths." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "`mobilenet_v1` and `vision_encoder_decoder` are unrelated architectures; same refactor pattern is not enough." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "Different model families (`mobilenet_v1` vs `speech_encoder_decoder`), so this is not one underlying change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "Both modernize output handling, but `vision_text_dual_encoder` and `univnet` are separate concrete implementations." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Same broad refactor style, but `vision_text_dual_encoder` and `upernet` do not share a single code-path fix." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "`vision_text_dual_encoder` vs Mamba/FalconMamba: different subsystems, so not duplicates." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Same output-tracing theme, but these are separate model refactors with different forward signatures." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "`univnet` and `upernet` are unrelated; the overlap is only in output-tracing cleanup." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model code paths (`univnet` vs Mamba/FalconMamba); not the same bug or fix." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "Both are tracing refactors, but `univnet` and `depth_anything` are distinct model implementations." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "`univnet` and `codegen` target different forward logic; shared theme only, no duplicate change." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "`upernet` and `rwkv` are unrelated model paths; the PRs are not the same concrete fix." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model families (`upernet` vs Mamba/FalconMamba), so these are separate refactors." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Mamba/FalconMamba output tracing vs `codegen` cache/output handling are different code paths." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "`timm_backbone` and `codegen` are unrelated implementations; same refactor pattern does not make them duplicates." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "`univnet` and `codegen` differ in model architecture and forward semantics." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "`upernet` and `codegen` touch different subsystems and are not one mergeable change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "`vision_text_dual_encoder` and `codegen` are separate code paths; only the refactor style matches." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Both are output-tracing refactors, but `depth_anything` and `vision_encoder_decoder` are different model families." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "`depth_anything` and `speech_encoder_decoder` are unrelated implementations, so not the same underlying change." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "`depth_anything` vs `superpoint`: different model code paths with only thematic overlap." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "`depth_anything` and `rwkv` are unrelated; same output-tracing cleanup pattern only." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "`depth_anything` and `dpr` target different architectures and cannot be merged as one PR." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "`vision_encoder_decoder` and `rwkv` share only the tracing-refactor theme, not a concrete bug fix." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "`vision_encoder_decoder` and `speech_encoder_decoder` are related in naming, but still separate code paths and PRs." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "`vision_encoder_decoder` and `superpoint` are different model implementations, so this is not duplicate work." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "`vision_encoder_decoder` and `dpr` are distinct forward paths; same general refactor style only." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "`speech_encoder_decoder` and `superpoint` are unrelated model changes, despite similar output-tracing cleanup." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "`speech_encoder_decoder` and `rwkv` affect different code paths and are not the same concrete change." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "`speech_encoder_decoder` and `dpr` are separate models; no shared duplicate bug is evident." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "`superpoint` and `rwkv` are unrelated implementations; same refactor theme is insufficient." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "`superpoint` and `dpr` are different code paths, so they should remain separate PRs." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "`rwkv` and `dpr` are distinct model changes with no shared concrete 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": 24, + "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-44662-3", + "nodes": [ + "pull_request:44662", + "pull_request:45075", + "pull_request:45401" + ], + "soft_pairs": [ + "pull_request:44662|pull_request:45401", + "pull_request:44662|pull_request:45075", + "pull_request:45075|pull_request:45401" + ], + "prepared_review_unit_hash": "b968fb6bc2fe42f3ed81c540ecec7177b3978c1b75a1d2c84ef7dff5e2b1ef2b", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 11071, + "estimated_input_tokens": 2768, + "estimated_eval_tokens": 5792 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 11071, + "estimated_input_tokens": 2768, + "estimated_eval_tokens": 5792 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three separate model-addition PRs for different architectures (PenguinVL, DeepSeek-OCR-2, Voxtral TTS). The shared file patterns are just Transformers boilerplate, not the same underlying change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical duplicate PR here: each PR implements a different model and code path, so they should stay separate.", + "best_issue_reason": null, + "best_pr_reason": "PR 45075 is the strongest standalone candidate: it is non-draft, has a full model integration including conversion tooling and tests, and appears more complete than the draft Voxtral TTS PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Different models and tasks (PenguinVL vs Voxtral TTS); the overlap is only in shared Transformers scaffolding and docs/auto-registry files." + }, + { + "left": "pull_request:44662", + "right": "pull_request:45075", + "accept": false, + "reason": "Both add unrelated model implementations for different architectures; shared filenames are generic integration points, not evidence of the same bug or change." + }, + { + "left": "pull_request:45075", + "right": "pull_request:45401", + "accept": false, + "reason": "DeepSeek-OCR-2 and Voxtral TTS are distinct model additions with different code paths and assets; they are not plausibly mergeable 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": 24, + "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": "740406d4aba4a9eef4c26cdcd9008e7364af201a307f74c44f2e12bbd9ce2d47", + "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": "Both pull requests are about fusion, but they target different layers: one introduces a general Module Fusion API, while the other wires up n-to-1 kernel fusion through KernelConfig and related integration changes. They are thematically related but not the same underlying change.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "Choose PR 45363 as the representative PR because it spans the actual end-to-end kernel-fusion integration path across core modules, config, and an example, whereas PR 44979 is a narrower standalone API addition.", + "best_issue_reason": null, + "best_pr_reason": "PR 45363 is the strongest cluster representative: it is a concrete, multi-file implementation of kernel fusion behavior and is more complete than the standalone Module Fusion API proposal in PR 44979.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979 (Module Fusion API)", + "right": "pull_request:45363 (n-to-1 kernel fusion via `KernelConfig`)", + "accept": false, + "reason": "Related fusion infrastructure, but different artifacts and different concrete changes: a new module-fusion API versus kernel-fusion integration and example code. They do not look 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": 20, + "component_count": 24, + "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": "bcadfdf72acceb8dc1cfbea34780e84b56ef5305fe9e4f1c12fb4c06e20a01aa", + "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 pull requests 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 tests and has clearer context.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45317 is the stronger canonical choice: it addresses the same bug as 45086 but includes test coverage and a fuller implementation description, making it the more complete fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 45317 is the best representative of the cluster because it fixes the concrete Mistral pre-tokenizer AttributeError and verifies the behavior with tests.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying bug and same code path in `src/transformers/tokenization_utils_tokenizers.py`; 45317 is essentially an expanded version of 45086 with tests, so these are duplicate fixes that could plausibly 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": 21, + "component_count": 24, + "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": "2585854094ac5783b4115ee2ded38e437fd8e270b30c53381b74f1b7f65c5aa7", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6137, + "estimated_input_tokens": 1535, + "estimated_eval_tokens": 3326 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6137, + "estimated_input_tokens": 1535, + "estimated_eval_tokens": 3326 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs work on Nemotron-H, but they target different concrete changes: #45115 is a narrower refactor around inheriting GraniteMoeHybrid and conversion/key renaming, while #45569 is a broader model-family split that adds dense/sparse variants, auto-mapping, config updates, and tests. They are related but not duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "#45569 is the better representative PR for this cluster because it is the more complete, newer implementation of the Nemotron-H family split and includes the full set of codepaths and tests, whereas #45115 is a narrower refactor focused mainly on inheritance and checkpoint conversion naming.", + "best_issue_reason": null, + "best_pr_reason": "#45569 is the strongest standalone PR: it covers the actual model/configuration registration work for Nemotron-H dense and sparse variants across auto mappings and tests. #45115 is useful background but is too narrow to serve as the main cluster representative.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Related Nemotron-H work, but not the same underlying bug/change. #45115 is a focused GraniteMoeHybrid-style refactor and conversion-mapping rename, while #45569 introduces separate dense/sparse model variants, config enum updates, auto-class wiring, and tests. These would not 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": 22, + "component_count": 24, + "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-45332-2", + "nodes": [ + "pull_request:45332", + "pull_request:45333" + ], + "soft_pairs": [ + "pull_request:45332|pull_request:45333" + ], + "prepared_review_unit_hash": "06c45e7a2683c407ff688dd9a6a0c90e4a0ff32effbc9fccd9f68e1b2b04c6ea", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5398, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5398, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs target heterogeneous-layer support, but they are not the same change: #45333 is config-only, while #45332 extends the feature into cache/modeling paths and multiple model files. Related, but not duplicate enough to merge as one PR.", + "confidence": 0.86, + "canonical_issue_reason": null, + "canonical_pr_reason": "#45332 is the broader end-to-end implementation of heterogeneous model support, covering config plus modeling/cache integration, so it is the better representative PR for the cluster.", + "best_issue_reason": null, + "best_pr_reason": "#45332 is the more complete and central change; #45333 looks like a narrower prerequisite/subfeature rather than a duplicate of the full implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "They share heterogeneous-config code, but #45333 only adds per-layer config while #45332 also changes cache/modeling logic and model-specific files. Same feature area, not the same concrete change or code path." + } + ] + }, + "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": 24, + "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": "45b9a731de260927adf80c4322c007a33d08ee64acba4a009b3319e61400c0cd", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5306, + "estimated_input_tokens": 1327, + "estimated_eval_tokens": 2910 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5306, + "estimated_input_tokens": 1327, + "estimated_eval_tokens": 2910 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are typing-focused maintenance, but they target different scopes: one is a broad type-checking sweep across many transformers modules, while the other adds specific typing support around modeling_utils/PEFT integration. They are related, but not the same underlying change.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45425 is the more focused and self-contained typing change, centered on concrete protocol/alias additions and modeling_utils integration; PR #45415 is a broad repo-wide type-checking pass.", + "best_issue_reason": null, + "best_pr_reason": "PR #45425 is the best representative of the cluster because it is narrower and more specific, making it a cleaner exemplar than the wide-ranging type-checking sweep in #45415.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Both are typing-related, but they address different changes: #45415 is a broad type-checking cleanup across many files, while #45425 adds focused typing support for modeling_utils/PEFT. They do not look mergeable as one concrete 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": 24, + "component_count": 24, + "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-45565-2", + "nodes": [ + "pull_request:45565", + "pull_request:45575" + ], + "soft_pairs": [ + "pull_request:45565|pull_request:45575" + ], + "prepared_review_unit_hash": "32f68244295d4371bc1aa86f3721e6c39c2b73a85a273a0be48febffd38798d8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4253, + "estimated_input_tokens": 1064, + "estimated_eval_tokens": 2384 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4253, + "estimated_input_tokens": 1064, + "estimated_eval_tokens": 2384 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs patch the same continuous-batching warning block in `src/transformers/generation/utils.py` and address the stale `num_return_sequences` warning; they look like two variants of the same fix, with 45575 being the narrower direct edit.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "45575 is the cleaner, more directly scoped fix: it removes the stale `num_return_sequences` warning and leaves the remaining `num_beams` check/message aligned with the code path.", + "best_issue_reason": null, + "best_pr_reason": "45575 best represents the cluster because it most precisely matches the underlying warning bug with minimal code churn.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45565", + "right": "pull_request:45575", + "accept": true, + "reason": "Same file, same `generate()` continuous-batching branch, same stale warning problem; the two diffs are close enough that they could plausibly be merged into 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/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/manifest.json b/snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..e0b158117e9aa5a56da9fe3ad11992bfddb066fe --- /dev/null +++ b/snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260422t120052z", + "artifacts": { + "hybrid": "snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-04-22T12:09:14Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260422T120052Z", + "variant": "hybrid" +} diff --git a/snapshots/20260422T120052Z/manifest.json b/snapshots/20260422T120052Z/manifest.json index ebbad94e9a05624b2434e45ecbd326127219bce3..825a7048e50376b9ad39e30c6c8264d91e17911d 100644 --- a/snapshots/20260422T120052Z/manifest.json +++ b/snapshots/20260422T120052Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 1787 }, "extracted_at": "2026-04-22T12:00:52Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260422t120052z", + "runs": { + "hybrid-model-20260422t120052z": { + "analysis_id": "hybrid-model-20260422t120052z", + "artifacts": { + "hybrid": "snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260422T120052Z/analysis-runs/hybrid-model-20260422t120052z/manifest.json", + "model": null, + "published_at": "2026-04-22T12:09:14Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260422T120052Z", "watermark": {